BART train-control computer cost-overruns
Adrian Brandt
(Published on May 26, 1998 in the Contra Costa Times)
Cost is up for BART computer
* Train-control system will run $23 million more and 2 years later
than planned
OVERRUNS
* April 1992: $36.9 million contract with General Railway Signal
Corp. to design, furnish and install a train-control computer
system. BART had estimated the cost at $75.4 million.
* July 1995: $1.7 million in additional payments, including
relocation of equipment buildings on the Dublin-Pleasanton line.
* December 1995: $410,000 due to weather delays at the
Pittsburg-Bay Point station.
* February 1996: $1.6 million because of a 240-day rain delay on
the Dublin-Pleasanton line.
* April 1997: $1.4 million for training, safety analyses and
equipment modifications.
* April 1998: $2 million because of complications on the
Pittsburg-Antioch and Dublin-Pleasanton extensions, including
water seepage into an equipment control room.
* Summer 1998: Contract due for final settlement. Total cost is
$60 million and expected to increase.
By Robert Oakes
TIMES STAFF WRITER
-----------------------------------------------------------------------
OAKLAND -- A computer system that controls BART trains zooming up to
80 mph on East Bay extensions will cost at least $23 million more
than expected and be completed two years late, according to transit
district records, BART officials and a contractor.
BART board members this summer could approve a final settlement of the
train-control contract, originally scheduled to cost $36.9 million.
The board in April approved payments pushing the price tag to $60
million, and the total will certainly climb due to rain delays,
construction problems and software glitches.
Agency officials said they're making the best deal they can without
risking lawsuits and an extended, costly court fight. Despite initial
problems, the computer system performs well.
"Some things were unavoidable," said BART Director Joel Keller of
Antioch. His district includes the North Concord and Pittsburg-Bay
Point stations, which operate on the train-control equipment.
"That's not to excuse the high cost," Keller said. "The public will
have a hard time understanding why the contract will come in so high
over the original cost."
The computer network along tracks and at stations oversees train
acceleration, braking and other movements, up to a maximum 80 mph.
With 1,000 or more passengers crowded into a rush-hour train,
equipment must function perfectly to prevent accidents and delays.
Critics said the contract is an example of how BART might have trouble
controlling costs on the next megaproject -- a $1.17 billion line
planned to the San Francisco International Airport.
"Who's going to pay the bailout on this project?" asked Bruce
Balshone, executive director of the Coalition for a One-Stop Terminal.
The Peninsula-based group is fighting the airport extension, which
opponents consider overpriced and unnecessary.
BART and the train-control contractor, General Railway Signal Corp.,
contend their deal is a unique case limited to the other extensions
in Contra Costa, Alameda and San Mateo counties.
"I think it just played out as well as it could have," said Dan
Donatello, a vice president with the Rochester, N.Y.-based company.
BART in 1992 estimated the equipment would cost $75.4 million, but
General Railway Signal submitted the low bid of $36.9 million. BART
in nearly all cases must give contracts to low bidders.
Two other firms submitted offers of $61.9 million and $104.9 million,
respectively, to do the same job. The General Railway Signal system
is completed, but some company personnel remain in the Bay Area to
oversee training, maintenance and warranties.
"We have a system from a well-established firm that specialized in
train controls," said BART spokesman Mike Healy. "Even with the change
orders, we've gotten probably the best control system we could have."
BART will conduct an internal audit to review all aspects of the
contract, said Director Pete Snyder of Dublin.
"I'm satisfied with the work GRS has done," said Snyder, who like
Keller was elected to BART after the original contract was approved.
"We have a system that is operating the way we want it to operate."
The core BART system built in the early 1970s operates on old
technology, and BART decided to upgrade when building the
Pittsburg-Bay Point, Dublin-Pleasanton and Colma extensions. The lines
opened from late 1995 to mid-1997.
What slowed the contract and increased costs? According to BART
documents, problems included:
* Weather. Rainy winters caused delays, including an eight-month
period lost to bad weather in 1994-95 on the Dublin-Pleasanton
line.
Train-control computers are among the last equipment installed, and
rain created a domino effect, delaying prior construction and
preventing the computer company from getting access to tracks.
At one point, water seeped into a train-control room and threatened to
short out equipment. BART agreed to pay the contractor for such extra
expenses.
* BART-ordered changes. Transit officials repeatedly requested
modifications to the contract, requiring General Railway Signal
to do extra work. BART has also agreed to pay for those expenses.
* Software. Connecting new systems with 1970s-era BART computers
proved far more complex than expected.
The train-control contract was originally scheduled for about 4.5
years but will last at least 6.5 years.
General Railway Signal still has some outstanding claims against BART,
and the contract can't be closed until the two sides negotiate a last
payment.
"That will be the final change order, and that will be it," Healy
said.
--
Adrian Brandt
(408) 565-7291 / a...@nt.com
Brian Mueller
Adrian Brandt, apparently paraphrasing from the Contra Costa Times,
wrote:
>Cost is up for BART computer
>
> * Train-control system will run $23 million more and 2 years later
> than planned
And no, it's not a waste of money. Yes, BART in its current form is
superior to caltrain in its current form.
----
Brian Mueller
E-mail: mulder78 at ix dot netcom dot com
20 year old male, in Morgan Hill, California, USA
(ten miles south of San Jose, California)
Erik Alm
Anyone know if the new BART train-control computer is Y2K compliant? Or
will they have to scramble to find money to fund that after the system
is on-line? The commute on Monday Jan 3, 2000 could be a lot more
difficult ...
James D. Umbach
The cat walked across Erik Alm <me...@metrodynamics.com>'s keyboard
and wrote:
Yes, it is nice that 1JAN00 will occur on a Saturday. That will give
companies two days to get stuff sorted out before the work week
starts. Of course, in my cynical nature, I really think it's
_ALREADY_ too late to get this problem licked.
---------------------------------------------------------------------------
| James D. Umbach | apostle (at) mother.com |
| Citrus Heights, California | my web site: http://www.mother.com/~apostle |
---------------------------------------------------------------------------
David Kaye
Brian Mueller wrote the quoted material below:
" And no, it's not a waste of money. Yes, BART in its current form is
" superior to caltrain in its current form.
In what ways is it superior? Is BART faster? Maybe so, maybe not. Is
BART friendlier to bicyclists? No. Does BART have cleaner cars? No.
Does BART have more service? Yes. Does BART have more comfortable
stations? Yes (especially compared to the non-existent Caltrain 22nd
Street Station). But, is a $23 million computer system justified? Well,
for a train that is forced to rely on radio signals and which cars
"disappear" when the tracks get wet or hot, maybe so.
Gosh, I hope they'll take $1 mill or so out of that money and maybe hose
down a few of their SF stations.
--
(C) 1998 Yes you CAN do Zen #29: To walk a trail in the forest
David Kaye and erase every sign of being there is a Zen practice
dk at wco.com
David Kaye
Erik Alm wrote the quoted material below:
" Anyone know if the new BART train-control computer is Y2K compliant? Or
" will they have to scramble to find money to fund that after the system
" is on-line? The commute on Monday Jan 3, 2000 could be a lot more
" difficult ...
Does BART need to do more than schedule trains based on a 7 day week? I
mean, is there anything special about any particular date of the month or
month of the year that BART's computer needs to keep track of?
--
(C) 1998 Yes you CAN do Zen #25: The act of Zen is not
David Kaye a process of doing; it's a process of undoing
dk at wco.com
Ron Newman
In article <6klis4$2gl$3...@news.ncal.verio.com>, David Kaye
<d...@removeme.wco.com> wrote:
> Erik Alm wrote the quoted material below:
> " Anyone know if the new BART train-control computer is Y2K compliant? Or
> " will they have to scramble to find money to fund that after the system
> " is on-line? The commute on Monday Jan 3, 2000 could be a lot more
> " difficult ...
>
> Does BART need to do more than schedule trains based on a 7 day week? I
> mean, is there anything special about any particular date of the month or
> month of the year that BART's computer needs to keep track of?
Well, there are presumably holidays on which service is to be run
on a Saturday or Sunday schedule rather than a weekday schedule.
--
Ron Newman rne...@thecia.net
http://www2.thecia.net/users/rnewman/
Scott Hazen Mueller
>Does BART need to do more than schedule trains based on a 7 day week? I
>mean, is there anything special about any particular date of the month or
>month of the year that BART's computer needs to keep track of?
BART may or may not have a year-dependent need (but they should be aware that
there is a February 29, 2000, should they not?) but it is highly likely that
any microcontrollers bought after around 1980 or so that have built-in
calendar functions (i.e. a lot of them) will have Y2K bugs. Switching, power
transmission and regulation, destination signs, and who knows what else will
all be vulnerable to unsuspected failures.
\scott
Bloody Viking
Erik Alm (me...@metrodynamics.com) wrote:
: Anyone know if the new BART train-control computer is Y2K compliant? Or
: will they have to scramble to find money to fund that after the system
: is on-line? The commute on Monday Jan 3, 2000 could be a lot more
: difficult ...
You'll find out soon enough! In reality, it's probably unknowable with
that embedded system. It's sort of like predicting which cars won't start
after the New Year's party.
Does the BART train "car computers" have some date-related function? If
not, it will likely work on Jan 3, 2000. However, you never know until
it's too late!
c.s.y2k added to crosspost.
--
CAUTION: Email Spam Killer in use. Leave this line in your reply! 152680
A cult founder's rustbucket freighter is his battleship.
2588516 bytes of spam mail deleted. http://www.wwa.com/~nospam/
Bloody Viking
David Kaye (d...@removeme.wco.com) wrote:
: Does BART need to do more than schedule trains based on a 7 day week? I
: mean, is there anything special about any particular date of the month or
: month of the year that BART's computer needs to keep track of?
Careful, there's holidays. If the train's "car computer" keeps track of
the week, it will have to know holidays, so there's a real danger of it
not being y2k-compliant. Worse, only the manufacturer of the "car
computer" has access to the sourcecode.
David Kaye
Bloody Viking wrote the quoted material below:
" Careful, there's holidays. If the train's "car computer" keeps track of
" the week, it will have to know holidays, so there's a real danger of it
" not being y2k-compliant. Worse, only the manufacturer of the "car
" computer" has access to the sourcecode.
But really, and someone help me on this: Don't the BART dispatchers or
some scheduling people somewhere determine how many cars are put into
service for what days? After all, extra cars are added for games at the
Coliseum, etc. So, this looks like a manual sort of scheduling thing for
me. So, then, why would BART even *need* a calendar in its train
computer? What would actually be controlled with a calendar?
--
(C) 1998 Yes you CAN do Zen #7: The only
David Kaye word you need to know is "look"
dk at wco.com
defr...@pacbell.net
> Erik Alm (me...@metrodynamics.com) wrote:
> : Anyone know if the new BART train-control computer is Y2K compliant? Or
> : will they have to scramble to find money to fund that after the system
> : is on-line? The commute on Monday Jan 3, 2000 could be a lot more
> : difficult ...
>
> Does the BART train "car computers" have some date-related function? If
> not, it will likely work on Jan 3, 2000. However, you never know until
> it's too late!
I am one of the Y2K Coordinators in my agency. We are going through enormous
steps to insure all of our programs and computer systems are Y2K compliant so
that on Jan 3, 2000 we don't have a rude awakening. This includes several
testing steps. I would hope BART is doing the same.
Edward M. DeFranco
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/ Now offering spam-free web-based newsreading
defr...@pacbell.net
In article <6kocl9$l46$2...@news.ncal.verio.com>,
David Kaye <d...@removeme.wco.com> wrote:
> But really, and someone help me on this: Don't the BART dispatchers or
> some scheduling people somewhere determine how many cars are put into
> service for what days? After all, extra cars are added for games at the
> Coliseum, etc. So, this looks like a manual sort of scheduling thing for
> me. So, then, why would BART even *need* a calendar in its train
> computer? What would actually be controlled with a calendar?
I would guess BART's computers are like the others I've seen in that the
operating system has a system date. If the computer is based on the
technology most are the date only sees the year in 2 digits. Therefore it
sees 99 as 1999 but 00 as 1900. In some computer we've tested, even
manufactured as late as 1991, when the system was tested with a yr of 00 it
would not boot back up. The error message was: incorrect yr in date file. We
We went through and had to reload the system from scratch, a several hour
ordeal. And the reload would only accept a more recent date (it was then 97).
We have retired (or are in the process of retiring) these systems. Even if
the date isn't necessary for the computer to perfom it's functions, if the
system will not boot up, nothing will work.
David K. Bryant
nos...@wwa.com (Bloody Viking) writes:
>Erik Alm (me...@metrodynamics.com) wrote:
>: Anyone know if the new BART train-control computer is Y2K compliant? Or
>: will they have to scramble to find money to fund that after the system
>: is on-line? The commute on Monday Jan 3, 2000 could be a lot more
>: difficult ...
>Does the BART train "car computers" have some date-related function? If
>not, it will likely work on Jan 3, 2000. However, you never know until
>it's too late!
Probably not the cars themselves. The BART system uses three
levels of semi-autonomous controllers. The one at BART Operations
(Lake Merritt) schedules trains, routes, etc. grants permission
to use sections of track, etc. That's Automated Train Operations (ATO).
It runs on Tandem and Sun(?). Then there's the Automated Train
Controller (ATC) [I may have them backwards]. That's what is on
the train itself. It takes the speed commands from ATO and controls
the motors & brakes, etc. to give the rapid accelleration and
glass-like braking we all know and love. Each station also is
involved in the process. The microscopic precision which enables
the train to stop within inches of the mark is part of the station
system.
Then there's the Automated Fare Collection (AFC) system. It knows
that you entered the system and jumped the gate getting out.
Bloody Viking
defr...@pacbell.net wrote:
: I am one of the Y2K Coordinators in my agency. We are going through enormous
: steps to insure all of our programs and computer systems are Y2K compliant so
: that on Jan 3, 2000 we don't have a rude awakening. This includes several
: testing steps. I would hope BART is doing the same.
That's the thing with y2k remediation. EVERYONE has to do their part to
keep civilisation going after the party. It's an "all hands" thing. :) I
wonder if BART is working on it. If not, and the trains do turn out to not
be y2k-compliant, the yuppies won't be able to get their coffee when all
the burger flippers can't get to work, and then that will cause more
accidents and work mishaps, and so on and so on...
Harlan Smith
David K. Bryant <dbr...@netcom.com> wrote in article
<dbryantE...@netcom.com>...
> Probably not the cars themselves. The BART system uses three
> levels of semi-autonomous controllers. The one at BART Operations
> (Lake Merritt) schedules trains, routes, etc. grants permission
> to use sections of track, etc. That's Automated Train Operations (ATO).
> It runs on Tandem and Sun(?). Then there's the Automated Train
> Controller (ATC) [I may have them backwards]. That's what is on
> the train itself. It takes the speed commands from ATO and controls
> the motors & brakes, etc. to give the rapid accelleration and
> glass-like braking we all know and love. Each station also is
> involved in the process. The microscopic precision which enables
> the train to stop within inches of the mark is part of the station
> system.
>
> Then there's the Automated Fare Collection (AFC) system. It knows
> that you entered the system and jumped the gate getting out.
So, since you seem to have some contact with BART, why don't you
investigate any possible Y2K sensitivity and report back to the group?
Harlan
David Kaye
Bloody Viking wrote the quoted material below:
" wonder if BART is working on it. If not, and the trains do turn out to not
" be y2k-compliant, the yuppies won't be able to get their coffee when all
" the burger flippers can't get to work, and then that will cause more
" accidents and work mishaps, and so on and so on...
What many people are unaware of is that in some computer operating systems
the calendar is simply patched and days added so that an internal 1900
is shown as 2000 with the proper number of days added so that 1/1/2000 is
on the correct day. What many designers didn't realize is that there's a
difference between the year 1900 and 2000: 1900 was not a leap year, 2000
is.
For today's lesson, everyone should set their computer's calendars to
2/28/2000, 11:59pm, then turn off their computers (assuming that this will
not affect any date-sensitive stuff on the computer). Then wait until the
date rolls over and then turn the computer on again. The computer will be
getting the date and time from the internal clock. Will it say 3/1/2000
or 2/29/2000?
--
(C) 1998 Yes you CAN do Zen #11: "The
David Kaye map is not the territory"
dk at wco.com
Alan Kline
David Kaye wrote:
>
> Bloody Viking wrote the quoted material below:
>
> " the week, it will have to know holidays, so there's a real danger of it
> " not being y2k-compliant. Worse, only the manufacturer of the "car
> " computer" has access to the sourcecode.
>
> some scheduling people somewhere determine how many cars are put into
> service for what days?
Absolutely...
> After all, extra cars are added for games at the Coliseum, etc.
Also correct.
> So, this looks like a manual sort of scheduling thing for
> me. So, then, why would BART even *need* a calendar in its train
> computer? What would actually be controlled with a calendar?
Probably, the only calendar-type function would be recalling the
day of the week, to load the proper schedules. Date and year wouldn't
be critical for this--that is to say, if 12/31/99 is a Friday, 1/1/2000
will inevitably be Saturday. It won't mystically jump to Tuesday, so
there probably is no reason for the ATO to track this. There must also
be some form of manual override, as well.
Actually, on 1/1/00 the SP ferries will magically reappear... ;-)
Seriously, it's a darned shame that we lost Harre Demoro many years too
soon. If he was still around, he and the Chronicle would be all over
any kind of BART/Y2K problems, and we'd all know about it. Not that
the Chron doesn't have capable people now, but I don't think anyone's
covered the transportation scene with quite the same passion...
Alan Kline
defr...@pacbell.net wrote:
> I would guess BART's computers are like the others I've seen in that the
> operating system has a system date. If the computer is based on the
> technology most are the date only sees the year in 2 digits. Therefore it
> sees 99 as 1999 but 00 as 1900. In some computer we've tested, even
> manufactured as late as 1991, when the system was tested with a yr of 00 it
> would not boot back up. The error message was: incorrect yr in date file. We
> We went through and had to reload the system from scratch, a several hour
> ordeal. And the reload would only accept a more recent date (it was then 97).
> We have retired (or are in the process of retiring) these systems. Even if
> the date isn't necessary for the computer to perfom it's functions, if the
> system will not boot up, nothing will work.
Point well taken. In my previous post, following-up to David, I was
thinking only of the application software, not the OS.
But, even if the ATO system won't boot, it's still possible to run
trains in road manual. Pain in the butt, and a lot slower, but at
least trains would move. The trains would still function, and as
long as the radios work, all is not lost.
Has anyone actually asked Mike Healy about this??
Alan Kline
James D. Umbach wrote:
> Yes, it is nice that 1JAN00 will occur on a Saturday. That will give
> companies two days to get stuff sorted out before the work week
> starts. Of course, in my cynical nature, I really think it's
> _ALREADY_ too late to get this problem licked.
Personally, I plan on cleaning out my bank accounts (won't take
long) and 401k on Friday, 12/31/99. Once the banks get their act
together, I'll put the money back... Or maybe not...my anniversary
is that weekend, so maybe I'd better take the crumbs and buy
something for my wife... ;-)
Bringing this back on topic, I didn't think that the ATO system
had that many date-critical functions. Does the wayside or onboard
gear really care what day it is, as long as there's the proper
separation between changes?
defr...@pacbell.net
> What many people are unaware of is that in some computer operating systems
> the calendar is simply patched and days added so that an internal 1900
> is shown as 2000 with the proper number of days added so that 1/1/2000 is
> on the correct day. What many designers didn't realize is that there's a
> difference between the year 1900 and 2000: 1900 was not a leap year, 2000
> is.
>
> For today's lesson, everyone should set their computer's calendars to
> 2/28/2000, 11:59pm, then turn off their computers (assuming that this will
> not affect any date-sensitive stuff on the computer). Then wait until the
> date rolls over and then turn the computer on again. The computer will be
> getting the date and time from the internal clock. Will it say 3/1/2000
> or 2/29/2000?
And if it works, try some of your software. Most times word processors will
work okay. But try a spreadsheet with formulas and calculations. Or a
database with date fields. Then you can really see if the software will
operate correctly.
David K. Bryant
"Harlan Smith" <hwsmith...@cris.com> writes:
>David K. Bryant <dbr...@netcom.com> wrote in article
><dbryantE...@netcom.com>...
>> Probably not the cars themselves.
>> Then there's the Automated Fare Collection (AFC) system.
>So, since you seem to have some contact with BART, why don't you
>investigate any possible Y2K sensitivity and report back to the group?
I wish I did. What I know I've learned from (1) reading about
BART, (2) noting the advertised positions, (3) talking with
train operators -- which is a stretch of a position description...
all they do is hold the doors open and talk on the PA system.
The computers do the rest.
Alan Kline
David K. Bryant wrote:
>
> I wish I did. What I know I've learned from (1) reading about
> BART, (2) noting the advertised positions, (3) talking with
> train operators -- which is a stretch of a position description...
> all they do is hold the doors open and talk on the PA system.
> The computers do the rest.
Far from it. This thread was beaten to death awhile back, so
I'm not going to waste a lot of time, but there is far more
to it than doors and PA. Particularly when things don't go
as they should.
Alan Kline
On Tue, 2 Jun 1998, David K. Bryant wrote:
> Alan Kline <akl...@nospam.crl.com> writes:
>
> >Far from it. This thread was beaten to death awhile back, so
> >I'm not going to waste a lot of time, but there is far more
> >to it than doors and PA.
>
>
> OK, they get to use the walkie-talkie and read the newspaper, too.
You wouldn't understand, so I'm not going to waste time trying to explain
it. Try DejaNews.
You must be a rocket scientist or something...
David K. Bryant
Steve Geller
On Tue, 26 May 1998 16:20:01 -0700, Adrian Brandt <a...@nt.com> wrote:
>Critics said the contract is an example of how BART might have trouble
>controlling costs on the next megaproject -- a $1.17 billion line
>planned to the San Francisco International Airport.
There's no question whether the BART-to-SFO-etc. project will over-run
(it already has over-run its committed funding). The question is
just how awesome an over-run it will be.
>"Who's going to pay the bailout on this project?" asked Bruce
We the taxpayers, of course -- probably as one of the reasons
Clinton's balanced budget will fail to happen.
> * BART-ordered changes. Transit officials repeatedly requested
> modifications to the contract, requiring General Railway Signal
> to do extra work. BART has also agreed to pay for those expenses.
This is the primary cause of software over-run, delay, and long-term
buggyness. It happens because the managers never spend enough time
thinking during the analysis and design stages.
> * Software. Connecting new systems with 1970s-era BART computers
> proved far more complex than expected.
That's baloney. The 1970s-era stuff is the hardware interfaces,
not the software. It shouuldn't be, anyway; it might be that some
managers tried to keep too much legacy code, which the rest of the
system had to be bent around.
>"That will be the final change order, and that will be it," Healy
>said.
Never say "final". <grin>
Stephen Oakes
Has a Simpsons joke appeared in this thread yet?
--
Stephen Oakes ste...@dbce.csiro.au
CSIRO Division of Building, Construction and Engineering
PO Box 56, Highett, Victoria, 3190, Australia.
Tel: +61 3 92526000 Fax: +61 3 92526249
little miss stacey
Hi,
Just poking my head in to wave hello to Bloody Viking.
.s
> Bloody Viking wrote in message <6knhud$no3$1...@hirame.wwa.com>...
Bloody Viking
little miss stacey (jac...@ix.REMOVETHIS.netcom.com) wrote:
: Hi,
: Just poking my head in to wave hello to Bloody Viking.
No worries, mate! :) Thanks for the G'day.
--
CAUTION: Email Spam Killer in use. Leave this line in your reply! 152680
A cult founder's rustbucket freighter is his battleship.
2625321 bytes of spam mail deleted. http://www.wwa.com/~nospam/