Slashdot Mirror
Denver Airport Automated Baggage System Abandoned
cherylchase writes "Denver International Airport opened in 1995 with an ambitious fully automated baggage system: 26 miles of underground track, thousands of small gray carts, all controlled by a mainframe programmed for just in time delivery. But the system never worked well; bugs delayed the airport's opening for months (at $1M/day). The system has now been abandoned as a cost cutting measure." From the article: "Technology, too, has brought change. Back then, the big-brained mainframe doing it all from command central was the model of high tech. Today the very idea of it sounds like a cold-war-era relic, engineers say. Decentralization and mobile computing technology have taken over just about everything, allowing airlines, warehouse operators and shippers like FedEx to learn with just a few clicks the whereabouts of an item in motion, a feature that was supposed to be a chief strength of the baggage system."
I've been in airports all over the place, I would think 26 miles of track underground wouldn't speed up the process, especially if it is unmanned. I trust eyes on my luggage more than nobody knowing if it is really being moved or not. I've had luggage take forever in JFK airport, and the fastest was in another country!
Go ahead and Google Denver International Airport and look into some of the conspiracy-theories surrounding the building, murals, underground facilities, etc. It's pretty wierd stuff, interesting to say in the least.
Whether or not it's true, I don't know. You decide.
I agree. Actually, this sounds like exactly the kind of application that a big 'mainframe' would excel at - thousands of transactions per second as baggage is tracked by sensors along the way.
I'm not sure why the idea of a mainframe is 'cold-war-esque', since they are still at the centre of much of what we do today.
Just as the implementation of IBM's OS/360 forms part of the "history" section of many Computer Science texts, so the Denver Airport baggage system is fast becoming history. The big difference of course being, OS/360 was a spectacular success, wheras Denver was a catastrophic failure.
Writing this stuff up is fine and good, but I think it would be worthwhile to try to learn from it. What was done differently?
If folklore serves me correctly, IBM was not afraid to throw money at the problem. I seem to remember they put two separate teams on the problem and took the best from each, fully conscious that half the effort would be thrown away. They sank as much money on it as was required, and ultimately succeeded.
Denver probably ate many more Dollars than OS/360, though I wouldn't know. But:
Apparently, this last has become a self-fulfilling prophecy.
I work in software development for an airline. It's amazing how much of a megaproject a reservation system is proving to be these days, and how many past attempts have failed. That's why one of the world's major reservation systems still runs in assembler on an IBM mainframe.
I think we're talking over-engineering, Big Design Up Front, profiteering, and (attempted, far too late) price-gouging.
Either that, or the only way to make a very large project successful is to code it in Assembler on an IBM mainframe.
When one person suffers from a delusion, it is called insanity. When many people suffer from a delusion it is called Rel
"Automation always looks good on paper," said Veronica Stevenson, a lead baggage handler for United Airlines and president of the union local that represents United's 1,300 or so baggage handlers in Denver. "Sometimes you need real people."
A system that would have streamlined and reduced the need for union employees has been found to not be very good by those union employees? Shock and awe, gentlemen. Shock and awe.
Robots do exactly what you tell them to. It only damaged luggage if the luggage wasn't loaded onto the robot correctly, it only misplaced luggage if the robot was told to go to the wrong place.
Can you blame me for wondering if the failure of this system was not entirely because of technical reasons?
the article does not draw any correlation between mainframe programming, software, or the failures of the system. a major flaw, according to the article was that: "The whirring baggage carts, programmed to pick up and drop off bags in a perfectly coordinated ballet, often just tipped over and dumped their loads." it also speaks vaguely about some "lizard tongue conveyor" whose failures would hardly seems the domain of software development. the denver baggage system fiasco sounds more like a failure in regular engineering, or at best, robotics programming. i hardly see why mainframe architecture, or any piece of software code, should be blamed as the primary culprit.
I'm an automation systems engineer. I always find the failure of systems such as this very interesting. I've done firefighting operations on many jobs where they were on their way down the toilet. Most of the time failures are caused by only one weak area in a project.. usually it's mechanical design problems, or software (logic) problems. I have seen an instance once where it was a union sabotage problem. It was interesting how that particular line would run perfectly well on it's own during the weekends; but during the week it was a disaster. Since I spend most of my time writing automation logic and robot programs, I tend to get stuck with developing software workarounds for bad mechanical designs. The worst that I recall was a tread booker for a tire plant. It was one of the most crude machines I've ever worked on. My favourite part was a coupling that tended to slip; I was asked to put code in that 're-homed' the servo axis every few minutes automatically. I was paid by the hour; I'll software patch the hell out of bad mechanics if you want! I'm not sure what the problem(s) "really" were in this instance, but it's kind of sad; what airport will be brave enough to try it again?
The hardware and software itself are a classical example of engineering failure. In this world of "agile" and "xp" people don't want to acknowledge that, instead they'll blame it on "mainframes" or "old school this" or IBM or whatever. Bottom line is BAE did just about everything wrong. The technology can work but they let stupidity get in the way.
Other than the opening days a decade ago, the system has one of the best records of any US airport. The problem is that when trying to decide where to send it to, it would send it down the wrong ramp. basically, the system could not detect the bags correctly. Sadly, with RID coming online with baggage, the system would have been made reliable. Since this system will NOT be ripped out, it may still be brought back in the future.
I prefer the "u" in honour as it seems to be missing these days.
I was living in the area when DIA was being built (and life sucked badly after it was complete - pretty much everyone I talked to preferred Stapleton for many reasons, the simplest of which was that you didn't have to drive 5 miles at 25mph after getting your short-term parking ticket that charges by the tenth of an hour).
Anyhow, I remember they held a press conference when they finally started the baggage system, and it was one of the funniest things I've ever seen in my life. Suitcases were flying every which way, often ripped in half, and the reporters were all hitting the deck! Of course, this was funny to me because I wasn't down there dodging flying Samsonites; one of the problems with the baggage system was the startlingly high rate of Workers Compensation claims of the workers who had to deal with it, and the most-common cause of injury was, unsurprisingly, falling items.
If anyone has a link to that video, I'd love to see it again. I've tried, but no luck. Maybe some enterprising soul in one of the Denver local news channels can put it up on their website as part of the story of the system's closure?
political_news.c: warning: comparison is always true due to limited range of data type
Hardware Lies.
Which means, those laser scanners don't always read the label as they should, boxes and things get caught on edges and don't move even when the conveyor is on, electro-mechanical equipment doesn't always work, switches sometimes stick, etc, etc. Your job as a software engineer is to anticipate these and to try to make sense out of the information the hardware's giving it, even though something may be garbled, and write your program so that the system can keep running and that operators are made aware of the mechanical problems the software is seeing so they can correct the situation.
I'm posting anonymously, because I was a maintaince guy on the baggage system until a couple weeks ago when I decided I'd better look for somewhere else to be because it became painfully clear that my job was going away. I've been there for ten years, and I'll admit that the machine has had some problems... But it very rarely goes down to the point to dosen't work.
The fact is, it's *the biggest machine in the world*, bar none. There's almost 30 miles of completely automated track, more motors and linear accelerators than you can shake a stick at. A machine that big is going to require lots of time to shake out, and that happened about 5 years ago.. It's been running very smothly since then, because we've established protocols to cataloge and rank priority of repairs. You can't imagine the dynamic loads on it. 1/4 inch thick track pieces can snap in two if they weren't repaired correctly, and yeah, 5 years ago we were having problems. It's all but solved today, it's very smoth running and despite it's costs, it's STILL the cheapest way to move bags around in the world.
Those baggage handlers are full of shit, it should be known. One of the main reasons it eats baggage is because those asshats load the baggage like morons. I've seen more panties strewn about than you'd like to know, and it's almost always a women's bag that gets ate. You know why? They pack the fuckers like sausages, and the baggage handlers just plain don't load them correctly. They won't put them flat, one end will be hanging out--and it's always the heavy end. The machine has close tolerances in some places where tracks intersect go over-under, and tight turns that can fling the bags out if they're loaded poorly, I mean they're going 30mph at some points, an improperly loaded bag will get tossed, regardless of it's weight or size.
It is a mechanical monster, no doubt, any machine that big is bound to be... But it baffles me why they've got to shut it down at the peak of it's opperating efficeincy. It's never run so good, and they decide to kill it after we tamed the beast. You should realize that the command and control system they have in place that operates the machine is always being optomized, and sometimes poor programming has led to breakdowns and increased baggage eating.
Conveyors will be much less efficient, and the airport dosen't have the infrastructure in place to handle the entire load of bags by hand, and even if they did it will be far more expensive. There are 4 turnstyles that will need to be built soon--and airport construction is anything but fast.. Like I said, it dosen't make anything but political sense to shut the machine down.
Sure, deep blue and other contendant for the 1000000 googlyflop use mainframe for computer power. but most of the world do not use mainframe for that, but for transaction based system which handle a LOT of transaction per second with real time big database (for example airline RES system, bank system...) and security (transaction termin properly and start properly, and save data properly on disk). Processing power is your LEAST problem. Plus those mainframe system are so old they have been debugged by 2-3 generations of programmer 10 times over. Due to the size of the code and the constant debugging I doubt you can get this kind of quality on PC with a new software and a lot of investement...
C. Sagan : A demon haunted world:
http://www.amazon.com/gp/product/0345409469/
visit randi.org
With all of the cost overruns, the wierd artwork, and the abandoned baggage system, DIA is still the single most usable airport in the United States.
1: There is more room for security which leads to shorter lines. Additionally, connecting flights don't require going through security again, further decreasing the load.
2: The airport design is simple and easy to understand. There is only one terminal building to arrive at, and the concourses are arranged logically.
3: The terminal is very nice - well lit and refresingly open. There is a distinct "open air" feeling that doesn't exist in many airports. There is a wide range of services as well - plenty of food, bookstores, coffee, etc.
4: Unlike Stapleton, snow doesn't shut down DIA.
5: The train system is fast and effective.
6: There is room for expansion, which is particularly important as Frontier expands (DIA is a major United hub, and the only Frontier hub).
7: The large size of the airport and openness of the runways make it easier to land and eaiser to route traffic.
DIA is the world's 10th largest airport. Give it a bit of credit.