Slashdot Mirror
Why BART Is Falling Apart
HughPickens.com writes: Matthias Gafni writes in the San Jose Mercury News that the engineers who built BART, the rapid transit system serving the San Francisco Bay Area that started operation in 1972, used principles developed for the aerospace industry rather than tried-and-true rail standards. And that's the trouble. "Back when BART was created, (the designers) were absolutely determined to establish a new product, and they intended to export it around the world," says Rod Diridon. "They may have gotten a little ahead of themselves using new technology. Although it worked, it was extremely complex for the time period, and they never did export the equipment because it was so difficult for other countries to install and maintain." The Space Age innovations have made it more challenging for the transit agency to maintain the BART system from the beginning. Plus, the aging system was designed to move 100,000 people per week and now carries 430,000 a day, so the loss of even a single car gets magnified with crowded commutes, delays and bus bridges. For example, rather than stick to the standard rail track width of 4 feet, 8.5 inches, BART engineers debuted a 5-foot, 6-inch width track, a gauge that remains to this day almost exclusive to the system. Industry experts say the unique track width necessitates custom-made wheel sets, brake assemblies and track repair vehicles.
Another problem is the dearth of readily available replacement parts for BART's one-of-a-kind systems. Maintenance crews often scavenge parts from old, out-of-service cars to avoid lengthy waits for orders to come in; sometimes mechanics are forced to manufacture the equipment themselves. "Imagine a computer produced in 1972," says David Hardt. "No one is supporting that old equipment any longer, but those same microprocessors are what we have controlling our logic systems." Right now BART needs 100 thyristors at a total cost of $100,000. BART engineers said it could take 22 weeks to ship them to the San Francisco Bay Area to replace in BART's "C" cars, which make up the older cars in the fleet. Right now, the agency has none. Nick Josefowitz says it makes no sense to dwell on design decisions made a half-century ago. "I think we need to use what we have today and build off that, rather than fantasize what could have been done in the past. The BART system was state of the art when it was built, and now it's technologically obsolete and coming to the end of its useful life."
Another problem is the dearth of readily available replacement parts for BART's one-of-a-kind systems. Maintenance crews often scavenge parts from old, out-of-service cars to avoid lengthy waits for orders to come in; sometimes mechanics are forced to manufacture the equipment themselves. "Imagine a computer produced in 1972," says David Hardt. "No one is supporting that old equipment any longer, but those same microprocessors are what we have controlling our logic systems." Right now BART needs 100 thyristors at a total cost of $100,000. BART engineers said it could take 22 weeks to ship them to the San Francisco Bay Area to replace in BART's "C" cars, which make up the older cars in the fleet. Right now, the agency has none. Nick Josefowitz says it makes no sense to dwell on design decisions made a half-century ago. "I think we need to use what we have today and build off that, rather than fantasize what could have been done in the past. The BART system was state of the art when it was built, and now it's technologically obsolete and coming to the end of its useful life."
As one who rides Washington D.C.'s metro rail every day risking death by electrical fire, shooting and/or mugging I feel your pain.
Lack of money, lack of expertise, lack lack lack
I suspect BART and DC's Metro have similar problems (even though the funding sources are a little different)
Can eat my shorts.
The fact that something is old does not mean it's ipso facto obsolete or that its design principles haven't remained sound. Conversely, the fact that something just got posted on github yesterday and uses the latest node.js and boost libraries doesn't mean it's been well designed. These are very different things.
I've rarely ever taken the BART and don't live in the the Bay any more, let alone the San Francisco proper, but it'd be nice to have an analysis that doesn't conflate the two.
Hire a Linux system administrator, systems engineer,
it's 40 years old at this point which is about the time that most big transit projects need a lot of money to rebuild and upgrade the system
Also used in India. This could be foresight on the part of BARTs designers, as they anticipated accommodating increased ridership by placing passengers on top of the cars. The wider gauge is more stable and less likely to shake them off.
Have gnu, will travel.
That might buy you a couple of thyristors. Not enough to make a difference.
What this all boils down to is the age old problem of money being available for construction, not maintenance or improvement. Follow up costs are ALWAYS lowballed. At least in the military sector, they explicitly cost out spares and upgrades (or at least cost out some of it). In civilian government it's always the shiney. Once it's running, no more ribbon cutting ceremonies.
To be fair to the BART designers though, If I designed something that lasted twice a long as specced and carried four times the passenger load, I'd be pretty happy.
Faster! Faster! Faster would be better!
Not true, actually. Engineers (remember, trains don't have drivers) actually watch the track ahead of them and respond to various conditions, including animals on track, broken down trains on track, and, perhaps most importantly, idiots standing on the yellow tiles at the station. You've clearly never ridden... or you'd have some idea just how often the engineer has to stop short of the station while the station manager gets on the PA to tell people to get off the yellow tiles, while everyone else waiting to get on the train is deciding whether to pull them back from the track, or push them onto it for delaying the train. Engineers also respond to various issues with the train itself; for example, I was on a car that had a stuck brake once; it took the engineer one stop to determine what the problem was, another to determine which car, and a third to get the attitude of that car and the car on either side of it adjusted such that the affected car remained level while the affected wheel was lifted off the track enough to alleviate the risk of the brake spontaneously combusting without making the train unstable. Once that train reached the end of the line, the affected car was removed, but the engineer had to get it there, first. Even track switching isn't automated on the BART system, so the engineers do that as well.
APK quotes people (including myself) without context and should not be trusted. Just thought you should know.
Due to the volumes of documentation available, BART is the longest section in the book "Great Planning Disasters". But the failures are human and the disaster started with the initial lies. After authorization of the new district and system failed a couple times at the polls, it was finally approved at the ballot as a system that was promised to be fully funded by fare-box revenue. It was designed with the idea of maintaining San Francisco as the economic core of the Bay Area. And almost everything was non-standard. They assumed people would drive to nearby stations then transfer to BART. That didn't happen at the rates expected and they *still* have a severe lack of parking. They claim they are getting over 20-times the customers they originally predicted and they *still* can't cover costs.
When it couldn't be built on budget, a temporary 0.5% sales-tax was imposed throughout the district. When it couldn't even come close to covering costs from the fare-box, the tax became permanent. I now pay for BART through sales-tax, property-tax and various federal and state subsidies. Despite this, a couple years ago the BART directors claimed they had a "surplus" and reduced fares. This when the tracks howl due to insufficient maintenance and, obviously, things are falling apart.
BART has had 40 years to save and plan for maintenance and upgrades and has utterly and completely failed to do so. Now that they have suddenly figured out that stuff wears out, they want 3.5 billion more.
Answering critics of the California high-speed-rail projects a state politician responded, "they said that about BART in the beginning, too." I fear he is all too correct.
~~~~~~~
"You are not remembered for doing what is expected of you." - Atul Chitnis
To be fair to the BART designers though, If I designed something that lasted twice a long as specced and carried four times the passenger load, I'd be pretty happy.
Actually it is closer to 30 times the passenger load, TFS lists the original spec of 100k / week , with todays usage of 430k / day ... or ~3 million / week.
To err is human; effective mayhem requires the root password!
We have schools throughout the country that are asking for voter approval for huge bonds to upgrade or replace their aging schools.
One school district near me tries to get voter sympathy by giving tours of its boiler rooms and showcasing a 60 year old boiler (that still works BTW).
During one of the trips a person on the tour asked our tour guide "The boilers didn't become 60 years old overnight - why didn't the school board put some money away every year for future maintenance and upgrades?"
I suspect BART is also the victim of failing to plan for the future. Entropy always wins. No system exists that will not need maintenance or repair in the future. It is foolish to defer maintenance and upgrades and shows a lack of stewardship by the managers of that system.
To the surprise of no one - the $70 million bond request by the school district was voted down by a 3 to 1 measure.
The only thing that they really messed up was the track gauge, not using the standard gauge but a more or less unique for the BART system.
When it comes to electronics it's not impossible to replace. I don't see a $1000 price of a thyristor as something remarkable if it's a high power type.
What's more amazing is that in many cases processors designed in the 70's are still manufactured today while processor designs from the early 90's are almost unobtanium now. Just look for Z80 processors (a 70's design) at Mouser and then look for 386, 486 or 68040 processors. You can find the latter but only at more obscure vendors and sometimes they are refurbished from scrapped computers. With that as background I'd rather try to fix a computer board from '72 than one from '92 if I have the schematics. A '72 computer is either wired or hole-mounted 0.10" split DIL chips on a PCB with maybe 2 or at worst 4 layers. A '92 computer board is way tougher and requires patience since some chips aren't just surface-mount soldered but glued as well on a PCB with multiple layers.
If you want to make a system that is going to have a long lifespan, then you have to design it with a lot of standardized interfaces using connectors that are extremely common and that are easy to manufacture. And when you do that also make sure you document the interface very well, since that allows people in the future to manufacture plug-in replacement modules using modern hardware. The overall design will be more expensive but the concept lifespan will be a lot longer.
If builders built buildings the way programmers wrote programs, then the first woodpecker would destroy civilization.
And while you're stuck in a massive traffic jam because everybody else took your advice, you can smile smugly that you avoided the taint of communism.
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
Sometimes the road more frequently traveled is, in fact, the better road.
I run into this all the time. My company manufactures a wire harness product that uses several connectors which is used on an OEM auto that sells in large volumes. The engineers could have designed in pre-existing, widely available and standard connectors available from numerous sources for reasonable amounts of money. Instead they decided to custom design some new connectors for the application despite the fact that they provide zero extra functional benefit, cost substantially more, have 4 month lead times for delivery, have to be ordered in 50,000 piece quantities and can be purchased from precisely one source. Whichever engineer came up with this idiocy probably added their entire salary over the lifetime of the product in unnecessary cost to this product. (We sell about 250,000/year at around $4 each so it would be easy to get $100,000 in cost per year out of this product with a more sensible design)
The wiring harness industry is awash with countless different unnecessary designs of terminals, connectors and other hardware than never should have been seen the light of day. I have a bookshelf 10 feet from me as I type this that has probably 120 thick catalogs that are full of redundant, unnecessary or non-standard hardware. Maybe 5% of those designs are actually necessary and the rest are nothing but waste.
My basic take is that while there is nothing wrong with going bespoke in principle, you need to have a VERY good reason to deviate from standards or to use unusual designs, even if those standards aren't totally optimized for your application. Engineers who don't understand or ignore this principle are essentially engaging in a form of malpractice.
They need that many thyristors because there was a voltage spike that was killing them.
Rather than fix the voltage spike on that one small section, they took other cars from other areas of the system, and replaced the cards with the blown thyristors.
Which the unfixed voltage spike then killed up as well.
Rather than bus-bridge the impacted section, and actually figure out what the heck was going on with that small section that was making it cook thyristors in the cars, they ... you guessed it! Threw *MORE* cars at the problem, and cooked even *MORE* of them.
Either someone is grossly incompetent, or someone really wants the taxpayers to buy them new toys, and they are perfectly willing to set fire to the old toys they no longer want in order to temper-tantrum their way into the new toys.
Meanwhile: quit being assholes and throwing more of your dwindling supply of cars at that section of track!
---
Moral of this story...
Patient: "Doctor, it hurts when I do 'this'!!!"
Doctor: "Then don't do that."
The expected benefit was not blowing off the golden gate bridge. The BART was supposed to go to Marin. That is why the gauge is broad. It made good engineering sense based on the design requirements. Then Marin pulled out.