AirAsia crash & automation failure

[Tefal]

would you rather get on a plane and hear (in a stephen hawkings voice) hello, my name is i7 920 overclocked to 4 giga hertz, and il be your captain today
?

yes, because i'd know he had is 2-3 buddies there as back ups, and as far as im aware there hasn't been a recorded incident of all the computers falling asllep at the controls.

the same cant be said of the pilots.

 

[Tefal]

Do you not think the official report might be a better place to start? All we have at the moment is speculation - some of it informed speculation by qualified people but it's still speculation unless they have first hand sight of the evidence which as yet nobody outside of the ongoing investigation does.

one big problem with this"informed opinion" is it's informed by seconds hand information from unqualified journalists rushing to a deadline and trying to make what is incredibly boring information interesting for sale to the public.

like you say unless they've seen the investigation evidence they're not reliable at all because what they're basing their opinions on could be wildly mis reported.

 

[AJK]

To be honest, yes! Most plane crashes are caused by the pilot **** ups, not the machine failing.

And how many potential crashes are averted by pilots taking appropriate action when the machine fails?

 

[dannyjo22]

they use different software on each computer so there cant be a systemic fault across all three.

So you are saying for instance ELACs 1 and 2 have completely different software?

 

[Tefal]

So you are saying for instance ELACs 1 and 2 have completely different software?

http://ifs.host.cs.st-andrews.ac.uk/Resources/CaseStudies/Airbus/Airbus-fcs.pdf

Hardware
diversity
●
The primary and secondary flight control
computers use different processors.
●
The primary and secondary flight control
computers are designed and supplied by
different companies.
●
The processor chips for the different
computers are supplied by different
manufacturers

The software for the different channels in each
computer has been developed by different
teams using different programming languages.
●
The software for the primary and secondary
flight control computers has been developed
by different teams.
●
For the secondary computers, different
languages are again used for the different
channels in each machine

with 3 primary and 2 secondary computers and only one is needed so there's quintuple redundancy in the flight control computer.


not sure if it goes that deep but for the main flight control systems they're as different as they can make them to try and avoid any common fault bringing every system down.

 

[Jokester]

And how many potential crashes are averted by pilots taking appropriate action when the machine fails?

Not as many, it's not just the airline industry, any industry that relies on humans as a safety function show that time and again the weak point is the human being. Anything that can be done to remove human control should be done from a safety point of view.

 

[LockyOK]

Anything that can be done to remove human control should be done from a safety point of view.

Would you be happy then to drift off to sleep in the Google Self-Driving Car as it made its way from Carlisle to Manchester in the rush hour?

 

[Slam62]

Would you be happy then to drift off to sleep in the Google Self-Driving Car as it made its way from Carlisle to Manchester in the rush hour?

Sounds good to me

 

[Jokester]

Would you be happy then to drift off to sleep in the Google Self-Driving Car as it made its way from Carlisle to Manchester in the rush hour?

If it's been certified safe for the roads, why not? I would be more worried about the other drivers on the road.

 

[Uhtred]

you can't tell me that computers can do a worse a job at driving than humans. Have you not seen the idiots on the road? I can't wait for the day we are taken off the road, although I'm not sure that will ever happen.

 

[Tefal]

you can't tell me that computers can do a worse a job at driving than humans. Have you not seen the idiots on the road? I can't wait for the day we are taken off the road, although I'm not sure that will ever happen.

least a computer isnt going to be looking down at its phone drifting across the lines at 80 +

 

[grumpysculler]

for the actual fly by wire stuff to fail, 3 separate independently designed computers running 3 separate independently designed software's mush all fail.

they use different software on each computer so there cant be a systemic fault across all three.

In a basic triplex system, it only takes two failures to bring down the system (either the two faulty units outvote the healthy one or you get three different answers and direct law is invoked). It's been a while since we had those, though. Even the space shuttle had four primary computers and a completely independently built fifth backup.

Modern Airbus use 5 computers, each of which actually has two computers inside and any unit will cut itself off (fail passive) if the command computer disagrees with the partner monitor channel.

I believe that Boeing use 3 computers, each with 3 channels (one command, two monitors) with a similar approach.

At least one computer will be somewhere different on the aircraft - I think Airbus spread them out more than Boeing. This means that a fire or object strike doesn't take out all FBW capability. You then have to consider power supplies, cooling, etc to ensure all single point failures are removed.

 

[Tefal]

In a basic triplex system, it only takes two failures to bring down the system (either the two faulty units outvote the healthy one or you get three different answers and direct law is invoked). It's been a while since we had those, though. Even the space shuttle had four primary computers and a completely independently built fifth backup.

Modern Airbus use 5 computers, each of which actually has two computers inside and any unit will cut itself off (fail passive) if the command computer disagrees with the partner monitor channel.

I believe that Boeing use 3 computers, each with 3 channels (one command, two monitors) with a similar approach.

At least one computer will be somewhere different on the aircraft - I think Airbus spread them out more than Boeing. This means that a fire or object strike doesn't take out all FBW capability. You then have to consider power supplies, cooling, etc to ensure all single point failures are removed.

yeah i forgot aobut the two backups.

also the wiring and hydraulic circuits all take different routes, this came in useful when the 380 engine threw its rotor and severed a lot as it went through the wing., i think they split them up even more after that

 

[dannyjo22]

snip.

Oh ok I understand your post now, I was reading it like you was saying the software state of the Elac and Secs was different, which it isn't hence why they are interchangable. The different software comes in from different manufacturers of those boxes and differing software between the Com and Mon channel.

I understand what you are saying now, my fault. You are talking card level rather than box level software.

 

[Scuzi]

"Human error" as it is used today is a fallacy. It is a convenient sticking plaster, used by those who ironically make such trade-offs, as an attempt to explain a wider problem with a system which they don't understand.

When system architects and designers finally accept that an aspect of a socio-technical system is a human who acts in the context of the various demands and pressures on them, and not the human as imagined by the engineer, maybe we'll make leaps towards further resilience. Tentative steps have been made in that direction but there is a long way to go.

The answers will not be found by seeking out the failures.

 

[FoxEye]

you can't tell me that computers can do a worse a job at driving than humans. Have you not seen the idiots on the road? I can't wait for the day we are taken off the road, although I'm not sure that will ever happen.

You can almost guarantee that the first generation of auto-drive cars will have many, many more accidents than later generations. Not all failure modes will be known; not all possible circumstances will have been accounted for.

Designing aircraft has been a process of continual refinement, improvement, and better understanding.

I would not trust a 1st gen self-driving car with my life. No. It can have all the certification you like. It will still fail in unexpected ways.

 

[adolf hamster]

Surely the human pilot is just another flight computer in it's own way, with different hardware, programming and failure points to the others.

Perhaps if pilots spent more time actually flying their own planes they'd be better equipped to deal with computer failures especially in bad conditions. You cant expect someone to let a machine do his job for years and years then suddenly perform at peak effeciency when thrust straight in at the deep end when those machines fail.

 

[Blinkz]

There is a lot of rubbish floating around here. I've spent 5 years flying the A320, and I fly two Airbus wide bodies now. Before Airbus I flew a Boeing 737.

Firstly we don't know what caused the AirAsia crash and so until there is more information and a proper report from the investigation I think it is a waste of time to guess what happened or point fingers (which is never helpful anyway!)

Secondly, flying (or pretty much anything for that matter) will never be risk free. A pilots job is to minimise risk and make the flight as safe as is possible. Automation is an excellent tool, it helps in far more situations then it hinders. Replacing pilots with more computers is I think absolutely crazy. Computers by definition cannot think outside the box. It is impossible to think of everything that could cause an accident, and so it is impossible to program a computer to deal with all emergencies.

Any accident is a tragedy, however it is never useful to assign blame. An open reporting system is the most important part of minimising risk. Mistakes happen, to us all. We must learn from them and make sure we avoid them in the future.

 

[StriderX]

There is a lot of rubbish floating around here. I've spent 5 years flying the A320, and I fly two Airbus wide bodies now. Before Airbus I flew a Boeing 737.

Firstly we don't know what caused the AirAsia crash and so until there is more information and a proper report from the investigation I think it is a waste of time to guess what happened or point fingers (which is never helpful anyway!)

Secondly, flying (or pretty much anything for that matter) will never be risk free. A pilots job is to minimise risk and make the flight as safe as is possible. Automation is an excellent tool, it helps in far more situations then it hinders. Replacing pilots with more computers is I think absolutely crazy. Computers by definition cannot think outside the box. It is impossible to think of everything that could cause an accident, and so it is impossible to program a computer to deal with all emergencies.

Any accident is a tragedy, however it is never useful to assign blame. An open reporting system is the most important part of minimising risk. Mistakes happen, to us all. We must learn from them and make sure we avoid them in the future.

Its not impossible, its just not cheap enough and semi-autonomous aircraft haven't had the flight hours yet to learn from as far as i'm aware.

There's just far too much going on now for people to be present for much longer, ATC messages might as well be static half the time.

 

[Fuzz]

The most poignant point for me in that whole thread was the way the two are designed (airbus vs Boeing)

Boeing lets you flip a switch (or selection of switches) so you can take full and complete control of the aircraft.

Airbus doesn't!