However the catch 22 with this is that the amount of investment into
risk control is massive, even though in the longer term the 'sugary'
coating is that it gives a way to (at least in the short term) reduce
staff. Rail companies look at the cost savings by reducing staff as
well as turning 'human factor' risks into 'engineering' risks which
are much easier to quantity and create specifications for. Sometimes
that is almost impossibly difficult to achieve.
Of course the limits are that you cannot ever create an automated
system which can adequately react to every possible situation and
provide the same sort of outcome that would be expected of experienced
real human staff. Us humans can't either, but unless you give an
automatic system self-awareness and enable it to learn, there will
always be real human operators somewhere in the chain of control, even
if just to step in which things are 'outside the square'.
So driverless trains are a mixed bag and it suits some types of
operations but not all.
Craig.
--
http://www.flick.com/photos/navarzo3
Is the Bread Helga's? How Nomenic is your Nescio today?
> So driverless trains are a mixed bag and it suits some types of
> operations but not all.
Metro style operations are perfect candidates for full unattended ATO.
There is 30+ years experience with attended ATO. The engineering and
vital systems required are well understood. An underground Metro you have
have an ideally constrained environment to make unattended ATO a perfect
solution - and probably far better than manual control given the dull
working environment.
Our modern insane absolutely nothing may go wrong ever safety rules will
pretty much mean that a 'manual' driven metro line would have pretty much
the same (or even more to account for human error) safety systems than an
unattended ATO operation.
I would argue that for a new turnkey build, the cost difference
construction would be minimal. Unattended will win.
There are advantages in operation on a driver less line, especially when
it comes to needing to add extra services to cater for a demand spike.
Instead of chasing around trying to find qualified crew with hours
available, the OCC just instructs the computer to prep and send into
service an extra train or two. Depending on cold start preparation
procedures for a stabled train set you could have an extra train out on
the line and in service in minutes.
For example, Paris Metro line 14 which is unattended ATO operated, the
trains don't even have cabs, passengers can look out the front. Out of
the peaks when a portion of the fleet is stabled they keep a set on hot
standby at each end of the line. If the OCC notices (or is told) a
station is getting crowded, in minutes they can send a standby set into
service to increase the capacity on the route. On the other lines they
don't have this flexibility, as they have to find a driver, make sure
they have hours, get them out to the stabled train , prep , etc, etc...
Unattended operation works. It's well tested. It's more flexible and
it's safety record is excellent.
And the decision on ATO affects the infrastructure. London's Jubilee
line extension was built intended for ATO. Due to incredibly massive cost
overruns on the tunneling and the fancy (then new idea) moving block
signalling, The Jubilee Line extension is the most expensive per/km bit
of railway ever built, the ATO portion was canceled and conventional
block signalling installed in order to save money and get the whole
embarrassing mess into service.
There are numerous cross overs next to stations that can't actually be
used as the track layout doesn't allow the signalling to provide the
required 'overlaps'. The train following the one needing to use the cross
over has to be stopped so far back that it causes massive disruption to
the service frequency.
The cross overs had been sited on the assumption that the trains would
be unattended ATO, and that the automation would never let a train hit a
'train stop' at speed thus overlaps would not be required so they didn't
leave enough space for them.
British style overlaps after signals are a feature of NSW signalling,
but I'm not sure who else in Aus uses this philosophy. QR don't. But the
Jubilee line experience shows how it affects the infrastructure.
In the past month I've ridden two metro lines with full unattended
automation, and two running 'attended' automation. One of those 'attended
ATO' operations has been running since 1968. The 1968 logic is still
operating, a massive and complex project is currently under way to update
it. (How to cut over to a new signalling/automation system when you can't
close the line for 6 months to do it).
Technology is such now that I don't see any technical reason why a
Sydney suburban train couldn't run unattended ATO anywhere in the
suburban network. Track obstruction recognition would need to be added to
a 'standard' ATO equipment package. The technology to use a video camera
and laser range finder to detect and classify obstructions in front of
the vehicle exist and I've personally seen demonstrated on autonomous
ground vehicles. The people working on the systems now trust them enough
to walk in front of a rather heavy robot vehicle and watch it steer
around them.
Sounds very Victoria Line to me (London Underground).
>
> Technology is such now that I don't see any technical reason why a
> Sydney suburban train couldn't run unattended ATO anywhere in the
> suburban network. Track obstruction recognition would need to be added to
> a 'standard' ATO equipment package. The technology to use a video camera
> and laser range finder to detect and classify obstructions in front of
> the vehicle exist and I've personally seen demonstrated on autonomous
> ground vehicles. The people working on the systems now trust them enough
> to walk in front of a rather heavy robot vehicle and watch it steer
> around them.
>
Of course, DLR (Docklands Light Rail) operates as a driver-less light metro
under TfL (Transport for London) co-ordination from Bank through areas
immediately east of the City of London, through Docklands, north to
Stratford and across the River Thames . DLR runs elevated, on the surface
and underground (and for that matter, under water too). It is attended
inasmuch as a Passenger Service Officer rides the trains, and where
necessary, can manually operate a train from a control panel. The
passenger(s) sitting at the front need to temporarily vacate. I understand
that there have been times when the cars have not detected the electronic
signal to enter a platform - and these are the most common occurance of
manual intervention under Line Controller supervision.
It should be noted that DLR's computer update systems have an enormously
clumsy way of handling temporary speed restrictions for works (these have to
be manually re-entered each time the computers are rebooted: the scope for
error in this manual process is extreme. This process has led to accidents
and a slating of DLR by the Rail Safety authorities in the UK. It's amazing
that a database system holding the temporary speed and routeing changes
hadn't been overlaid, using a boot .exe programme to invoke the temporary
data over the standard data. Perhaps by now, it has.
DW out west
> It should be noted that DLR's computer update systems have an enormously
> clumsy way of handling temporary speed restrictions for works (these
> have to be manually re-entered each time the computers are rebooted: the
> scope for error in this manual process is extreme. This process has led
> to accidents and a slating of DLR by the Rail Safety authorities in the
> UK. It's amazing that a database system holding the temporary speed and
> routeing changes hadn't been overlaid, using a boot .exe programme to
> invoke the temporary data over the standard data. Perhaps by now, it
> has.
The DLR is on it's second (may be 3rd by now) ATO system. DLR trains are
not autonomous. After 40 years the Victoria line is just getting its 2nd.
The DLR trains are effectively remote controlled, the central computers
(the system is divided into separate regions) keep track of all trains in
their control area and send instructions to control them.
The original ATO was replaced as it couldn't cope with expansion. This
resulted in the original trains being replaced as it was too difficult to
fit the new ATO gear. (They sold them to a German city who fitted cabs
and run them as conventional LRVs).
The current control system also exceeded capacity, but it has a hand
over mechanism, so the DLR is really 3 separate systems now with
'interfaces'.
I can also see how they screwed up with the speed restrictions - in
normal operation the supervisory computer never restarts, total loss of
this system halts all trains (violently, they make emergency stops when
they lose the heart-beat from control) and takes several minutes to
restart and even longer to require the position of all trains and start
traffic. So I really can imagine some one saying 'this data does not need
to be saved to persistent storage'.
Meanwhile, back in the real world....
Victoria line's ATO is different and simpler, the local Signalling
determines the safe speed for the train and sends it via coded track
circuits. The OCC only know train positions by track occupancy as per any
conventional railway.
The Central line as a ATO where the train is told the distance to
obstruction (another train, station stop, etc) and it calculates the
speed curve required.
On both these lines, I assume drivers take manual control to deal with
speed restrictions. They would also have to take manual control to deal
with 'train regulation' issues that occur.
Neither line has platform doors either. People can and do fall onto the
tracks.
The Singapore one is interesting, the trains are autonomous - i.e.
unlike most automation systems where the train is driven from the ground
by an extension of the signalling logic, the Singapore trains really do
drive them selves. The on-board computer takes input from the signalling
system that is reporting 'distance to go' on a moving block
infrastructure. Based on this input, train location (that the train
determines itself based on the signals and a tachometer), it consults
it's own map for line speeds and it's timetable diagram and determines
what speed is appropriate for it to keep it's schedule with in the
constraints of line speed and signal clearance. Just like a human driver
has to do.
They had a 6 month delay bring in the line into operation - but it
wasn't the ATO part - the moving block signalling had issues. Their new
circle line based on the same technology seems to have quietly gone into
service with minimal issues. The circle line's problems have been civil.
(Tunneling collapse - and that bit is not yet open and way behind
schedule).
The Singapore NEL trains can be remotely driven from the OCC if required
and I think certain members of station staff are trained as drivers and
they can board trains and take manual control if required. And all the
remote operation equipment is duplicated so that if a computer failure
occurs and they lose automation redundancy they can move the train
remotely to a station and get a human down to deal with the situation.
As much as I'm enthusiastic about unattended train operation, the Sydney
Metro seems to be simply a political plot to stick it up Cityrail and the
railway Unions. We need western suburbs extensions to the heavy rail
network more than we need an inner city metro.
> The Jubilee Line extension is the most expensive per/km bit of railway
> ever built,
More expensive than "typical" italian stretches (e.g. Milan's Passante
?)
--
----------------------------------------------------------------------
nos...@mi.iasf.cnr.it is a newsreading account used by more persons to
avoid unwanted spam. Any mail returning to this address will be rejected.
Users can disclose their e-mail address in the article if they wish so.
> On Sun, 10 Oct 2009, Matthew Geier wrote:
>
>> The Jubilee Line extension is the most expensive per/km bit of railway
>> ever built,
>
> More expensive than "typical" italian stretches (e.g. Milan's Passante
> ?)
The table of costs I remember seeing is now several years old, but the
Jubilee line was well in front by a considerable margin. At the other end
of the scale was Madrid, in Spain who have managed to build Metro
extensions amazingly cheaply.
How does Jubilee extension compare with America's greatest
embarrassment, LA's main line.
The Wikipedia page says the Jubilee Line extension had a total cost of
GBP 5.3 billion. Other pages I found put the length at 10mi (16.1km),
which gives GBP 329 million/km.
NYC's Second Avenue Subway is projected to run cost USD 17 billion for
8.5mi (13.8km), or USD 1.2 billion/km, and who knows how much over that
budget they'll go before it's done...
I know the USD has fallen over the last decade, but I don't think the
exchange rate has reached 3.7:1 yet.
S
--
Stephen Sprunk "God does not play dice." --Albert Einstein
CCIE #3723 "God is an inveterate gambler, and He throws the
K5SSS dice at every possible opportunity." --Stephen Hawking
Supposedly the shibuya-ikebukuro section of the new Tokyo fukutoshin
subway line cost 250 billion yen (~ $ 2.5 billion), for 8.9 km of
deep-bored tunnel and 6 stations.
It seems surprisingly cheap to me though, so I always feel a little
suspicious of this figure (Japan's not exactly the cheapest place
generally).... but you see it all over the place.
-Miles
--
"She looks like the wax version of herself."
[Comment under a Paris Hilton fashion pic]
> Stephen Sprunk <ste...@sprunk.org> writes:
>>> The Jubilee Line extension is the most expensive per/km bit of railway
>>> ever built, ...
>>
>> The Wikipedia page says the Jubilee Line extension had a total cost of
>> GBP 5.3 billion. Other pages I found put the length at 10mi (16.1km),
>> which gives GBP 329 million/km.
>>
>> NYC's Second Avenue Subway is projected to run cost USD 17 billion for
>> 8.5mi (13.8km), or USD 1.2 billion/km, and who knows how much over that
>> budget they'll go before it's done...
>
> Supposedly the shibuya-ikebukuro section of the new Tokyo fukutoshin
> subway line cost 250 billion yen (~ $ 2.5 billion), for 8.9 km of
> deep-bored tunnel and 6 stations.
>
> It seems surprisingly cheap to me though, so I always feel a little
> suspicious of this figure (Japan's not exactly the cheapest place
> generally).... but you see it all over the place.
>
Does anyone have a breakdown for each on how much of the cost is land/air
space/tunnel boring costs, Vs the rest?
The replacement of the original vehicles was more to do with the
opening of the Bank extension. Because this is in single bore tunnel,
for safety reasons, the trains are requried to have end doors, which
the original vehciles did not. The doors on the originals were also
not particularly suitable for the much heavier than expected passenger
loading on the system.
The DLR has very much suffered over the years from having initially
been conceived of, and built, as a cheap and cheerful, low capacity
system that has had to cope with passenger demand more akin to a full
heavy rail metro line than its initial conception of Light Rail. The
number of times infrastructure that was practicallly brand new had to
be radically altered or abandonned altogether because the initial
design could not cope with the traffic levels is astounding (matched
by the transformation of the landscape through which the trains run,
which has similarly been transformed out of all recognition).
Robin
> The replacement of the original vehicles was more to do with the
> opening of the Bank extension. Because this is in single bore tunnel,
> for safety reasons, the trains are requried to have end doors, which
> the original vehciles did not. The doors on the originals were also
> not particularly suitable for the much heavier than expected passenger
> loading on the system.
Some recent photos of DLR at
http://www.flickr.com/photos/gunzel412/sets/72157620825361312/
--
---
Computer - vehicles for the mind; they'll drive you crazy
---
David Johnson
usenet.at.trainman.id.au
http://www.trainman.id.au
------------------------------------
These comments are made in a private
capacity and do not represent the
official view of RailCorp.