UK Controllers Say Air Traffic System 'Not Safe'

Jack Spine writes "Air traffic control technology being implemented in one of the major transport hubs in the UK is 'not safe,' according to air traffic controllers. The electronic flight data system (EFD) being phased in at Glasgow Prestwick Airport is too slow to handle real-time inputs, and could not cope with an outage that isolated it from the main air traffic system. Controllers had to scramble to handle the situation. Good luck if you're traveling to the UK anytime soon."

Comments

Been there before

[mbone]

I can remember when the US Air Traffic Controllers said that our ATC system was unsafe. Reagan fired them. (True, when they struck, but still...)

Re:Been there before

[drsquare]

Imagine the world if public sector unions had instead won this battle and felt free to impose their selfish wants on the rest of us. Scary, eh?

Yeah, America might have full employment, a better work-life balance, and higher incomes.

But I'm sure you're much better off after thirty years of Reaganomics...

Made the mistake.

[140Mandak262Jamuna]

I made the stupid mistake of actually reading the fine article. It seems to suggest that the old system of using paper strips and flight names is faster, more reliable and has bigger capacity to handle real time input than the new one based on computers and "smart stripes". Should have waited for people to read and post comments, that way some kind soul would have posted something easier to understand.

Re:Made the mistake.

[natehoy]

The phasing-in began on 28 January, but has suffered problems, including latency and screens not working, according to forum posts.

It may (or may not) take a controller longer to write a flight number down on a strip of paper and use that as their handoff device. But, once captured, handing off a flight is as simple as handing the strip to the controller you just transferred the flight to.. The strips on your board tell you which flights are yours to manage at the moment.

The computer should be able to do it faster and more efficiently and show you a nice list of flights on a screen, but if the computer starts slowing down or your screen fails, you're missing information you need RIGHT NOW to keep two planes from attempting to occupy the same airspace. Now you have to figure out which flights you are responsible for from memory, and whether flight 2345 has been handed off yet or not.

The old system is certainly less efficient than a computer. But you only need the system to be efficient enough to get you the information when you need it, no faster response time is needed. There's also a reliability component that is apparently not being met in this case. If the computer slows down (even occasionally) to a point that is slower than you need the information, or your screen fails, you're in trouble and so are the passengers on the flights in your zone of responsibility.

All things in aviation have multiple layers of safety backup. The paper-strip system is a backup / memory-aid for the controller. It helps the controller keep track of which flights are theirs, and it's pretty much not subject to failure. It's not fancy because it doesn't need to be, it just needs to work.

Trouble is, if you want to replace that with a computer, that's great, but the computer has to be as reliable as having strips of paper hanging about. So you need at least two computers at each station, one being a backup to the other, and even occasional slowdowns are unacceptable.

On 15 February, the IBM-based National Air Traffic Service system covering the UK stopped talking to EFD. Air traffic controllers at Prestwick scrambled to remedy the situation. Some people on days off had to go into work to try to move the traffic build-up, which caused numerous delays to flights

It states (not suggests) that the REAL problems came into the fore when the electronic communications that underlie the system failed. In other words, they had a system that was receiving no data, and had to fall back on the paper strip system, which they lacked the manpower on site to do effectively.

Not stated but probably also a problem was the fact that flights in progress were being tracked on the computer, and when those updates stopped coming in the controllers were actually missing information, so they had to quickly reconstruct everything on paper so they had a complete dataset to work with. Also not stated but probable was the possibility that manpower had already been reduced to take advantage of the efficiencies of the computer system, so they had to drag the people who ran the little strips of paper around back in to work in a hurry.

You can tape my punchtape when you pry it...

[davolfman]

Am I the only who seems to notice that the old system apparently ran off ticker tape somehow? WTF! How do you even make that work? If these people have been working on such a system that long no wonder they have trouble training them to a new system, it must all be reflex by now like driving a stick shift.

Re:You can tape my punchtape when you pry it...

[natehoy]

Not ticker tape. A Flight Progress strip http://en.wikipedia.org/wiki/Flight_progress_strip

"Old-school" controlling uses strips of paper. When a flight comes under your control, you grab a strip of paper and write the flight number on it. You stick it up on a board in front of you. When a flight gets near leaving your zone of control, you tell them to contact the next controller, and you physically hand the strip of paper to that controller or have a flunky run it down for you. Then when the pilot calls that controller, the controller is expecting the call (and if the pilot never calls the controller, the controller knows the pilot has screwed up his frequencies or made an error).

This is called "handing off" a flight, and handing the strip of paper over is the origination of that term.

And the problem here is not training the controllers to use the system. The problem is that the fancy new system had a tendency to slow down or fail, meaning the controllers needed to fall back on the strips of paper.

Re:"java"

[trollertron3000]

This can be true but a language doesn't design a bad system, a software engineer does. Many reliable systems have been built using Java and reside in hospitals, transportation, and power infrastructure. I can't blame a language or runtime for piss poor design. Also keep in mind not all Java applications or runtimes are the same.

Searching for more on this issue I found a post on how ATC insiders view it on the PPrune forums (UK site for professional pilots): http://www.pprune.org/atc-issues/427001-efd-scottish.html

Kind of an interesting behind the scenes look.

Re:Not just the UK

[natehoy]

But this isn't "old technology fails". This is a problem with the reliability of a NEW system they are trying to implement, and they had to fall back on the old system (handing around little strips of paper with flight numbers on them) in order to keep the operation running.

Technology is not constantly updated in ATC because controllers and pilots value reliability over new sexiness. So they tend to like to stick with what's worked well for years or decades rather than updating to the latest shiny every couple of years and trying to debug problems with it while flying about in metal tubes at hundreds of miles an hour with passengers onboard.

Call it paranoia if you like. I like the new shiny sexiness on my cell phone, but I want my instrument panel to be based on shit that works.

Re:Not just the UK

[ObsessiveMathsFreak]

You would think this is one of those places where the technology would be constantly updated, but not so.

Maybe you want the software which prevents your plane colliding with any of three dozen others written in php on a LAMP stack with automatic updates to your latest iDink app, but I don't.

Give me a 40 year old system written in COBOL any day of the week. At least there's a chance that was written by a Real Programmer(in TECO naturally).

Re:"java"

[LordFolken]

Java is the language taught to all these part-time it / project manager types of people. Its the only thing they know. This is why java has become a synonym for "Oh,OH, danger! Amateurs at work."

ATC Systems are highly political, i therefore doubt that the choice of language was up to the programmer. The project was probably cash-strapped, over budget, late etc. Now people are trying to force it down users throats.

In the forum you see a lot of discontent users, that reads to me like a classic project fail along the lines of: Boss:"We need to go digitial!" As to the how was never properly analyzed. And now people are stuck with a sub-standard system.

Its alarming to see this in the aviation industry. Its actually one of the more sane fields of it-work.

EFD does what exactly?

[destroyer661]

Can anyone comment on what exactly this system is doing? From reading TFA and googlin' around it seems like it needs to aggregate all(?) incoming/outgoing aircraft data into a readable strip that humans can then do something with? Please correct me if I'm wrong.

Re:"java"

[LWATCDR]

Actually you are wrong.
The Java haters are so funny. I have worked in Pascal, Fortran, Modula-2, c, c++, objective c, JAVA, BASIC, COBOL, PHP, Perl, and Python.
And for a desktop app that deals with a database I like Java the best. The problems you see with Java tend to be caused by the fact it is so easy to get a program to work even if you use terrible design. BTW yes I know about javascript and HTML5 but those are really new and frankly do not have the control and performance that Java gives a programmer IMHO.
I can see one reason for using this that you probably missed., portability. The US ATC system was so tightly designed that we where stuck with using old slow computers for years. The old system just couldn't be run on modern hardware.
Java should prevent that issue. As long as you can make a JVM work you are good. You could in theory use c++ and QT or Mono but I would consider those higher risk decisions than Java.

Re:The right tool for the job

[LWATCDR]

They are talking HUMAN realtime not microsecond jitter realtime.
You know like a text editor realtime.

Re:"java"

[KnownIssues]

That link to pprune.org was truly educational. You tend to have this idealized image of "professional" fields like police, doctors, lawyers... air traffic controllers. Reading this forum thread showed how things operate not one bit differently than anywhere else. Workers blaming management for pushing through a project without being fully informed. "Old-timers" lamenting the fall of the old system and its need for personal experience and its replacement with a more automated system that requires new users to be less trained. Bitterness about the new young pups coming in with their business degress and flow charts and company cars instead of being able to work the machine with their own two hands. Training given on a pending-obsolete version of the software just to keep project dates. And I'm only page 2 of 12!

It's nice to be reminded that humans are humans everywhere.

Safety is Not Binary

[eepok]

The air traffic control system is neither "safe" nor "not safe". It's impossible to be completely safe and the current system is definitely better than a worse system.

What they should say is, "... is not safe enough for X." where X can equal "the amount of money we put into it", "modern standards", or "the pilots and passengers".

Re:aye that's

[hairyfeet]

What I love is the picture in TFA. I mean could they get ANY more perfect when talking about a system being too slow than to show a couple of old folks outside the airport struggling with their luggage? Hell old folks struggling with baggage should be the picture under "too damned slow" in the dictionary!

As for TFA I'd love for someone to play "follow the money" and see whose palm got greased to pick this system. From the looks of it they picked a system used for BFE middle of nowhere airports and it simply doesn't scale well, which you'd think they would have wanted to see some realtime tests with the amount of data they are cranking before signing the check. My guess is someone there got a really nice "bonus" for picking this system over the others, safety be damned.

if they don't like it...

[ajdub]

the people who built it failed to do their jobs correctly. if there's one thing i can't stand, it's when technology is done wrong by people who don't know what they're doing, then foistered upon others by a heavy hand of management. if the system doesn't make the controllers happy, it's wrong. they're not whiney users... the system sucks.