ERTMS Regional and branch lines

[HSTEd]

Well the UIC and the Swedish railways have been pushing a low cost development of the ERTMS system for use on secondary routes with relatively low traffic levels which uses the equipment on a train designed for a conventional ERTMS implementation.

It effectively implements the signalling topology of the Level 3 proposals but does so without moving blocks, allowing to be implemented far more cheaply as there is far less work to do to produce an operational real-world system.

It allows the removal of the all trackside signals and almost all track circuitry, leaving axle counters or track circuits only in a limited set of locations and allowing the elimination of all signal wiring in favour of radio based communications systems.

They claim that this can reduce signal maintenance costs by upwards of 50%, which could have implications for the viability of various branch lines, especially ones that are currently operated by relatively self contained fleets of vehicles or where RETB is currently installed (and is nearing retirement).

So has anything been done by Network Rail in attempting to participate in this development programme, especially now the Cambrian Line ERTMS Level 2 appears to be functioning reasonably well?

 

[jopsuk]

In effect, is it essentially a 21st Century version of the RETB concept, with added benefits such as being able to run one train much closer behind another in the same section?

 

[HSTEd]

In effect, is it essentially a 21st Century version of the RETB concept, with added benefits such as being able to run one train much closer behind another in the same section?

Essentially, but the blocks are defined by the programming in the control computer.

 

[jopsuk]

and the system always knows exactly where the trains are, rather than just knowing which many-miles long stretch a train has a token for? Which would therefore presumably make it easier for signallers to give permission to cross for slow vehicles at crossings, and such matters.

 

[HSTEd]

and the system always knows exactly where the trains are, rather than just knowing which many-miles long stretch a train has a token for? Which would therefore presumably make it easier for signallers to give permission to cross for slow vehicles at crossings, and such matters.

Indeed, it can know the exact location of every train at all times.
However it needn't do so if there is nothing to control, this is what makes it cheap to implement because you don't actually need complete GSM-R coverage over the line, only over parts that have points or whatever.

 

[WatcherZero]

The UK by default, almost uniquely in Europe has effectivley already trialled ERTMS on rural lines. The problems encountered and the discovery theyve seriously underestimated how much radio spectrum is required means their unlikely to try even lighter.

 

[68000]

Can you explain? I understand the radio spectrum requirement is the same across Europe

 

[WatcherZero]

Theres the simple permissive information and radio but theyve discovered it takes up more bandwidth than expected, any further data exchange between the train and control centre (as is being proposed) would mean there wasnt enough radio spectrum in areas where lots of trains are in close proximity such as hubs. Therefore their thinking of allocating a second chunk of spectrum to ERTMS.

The additional functions their proposing communicating through GSM-R are CCTV video, cargo tracking, passenger information, etc...

 

[Nym]

"They're"

And why not just use the standard GPRS EDGE or 3G networks for those ancillary services, it isn't safety critical, so doesn't need the gurenteed connection offered by GSM-R...

 

[WatcherZero]

Because of the reception problems of moving vehicles and there can be stretches of track with little cells or non 3G cells. If your remotely monitoring the trains CCTV you dont want it to keep cutting out. Anyway all they have to do is allocate more radio spectrum and a backwards compatible update to the standards such that basic services would be in the existing allocation while additional services would be in the new spectrum.

 

[Railsigns]

"guaranteed"

 

[jopsuk]

Hehe, the inevitable fail of the spelling/grammar corrector. Anyway, Railsigns, presumably if the entire network was fully ERTMS 2 or above, your (rather good) website would become wholly a historical one?

 

[Railsigns]

Anyway, Railsigns, presumably if the entire network was fully ERTMS 2 or above, your (rather good) website would become wholly a historical one?

Going by the variety of signs still to be seen on the Cambrian Lines today, probably not.

 

[Gareth Marston]

Given the amount of kit needed for ertms and extra staff I've seen on the Cambrian as it stands ertms is backward move as far as costs are concerned for any branch/rural line. A modern RETB is whats needed not mile after mile of dodgy axle counters like we ave on the Cambrian now.

 

[HSTEd]

Given the amount of kit needed for ertms and extra staff I've seen on the Cambrian as it stands ertms is backward move as far as costs are concerned for any branch/rural line. A modern RETB is whats needed not mile after mile of dodgy axle counters like we ave on the Cambrian now.

ERTMS Regional would have no axle counters whatsoever.

It would identify the location of the train using Eurobalises and/or axle odometry aboard the train itself.

 

[Nym]

Or GPS...

 

[HSTEd]

Or GPS...

Indeed, the original ERTMS steering commitee came to the conclusion that it was not viable at the time, however the Americans have done a great deal of very interesting work on differential GPS in connection with Positive Train Control, so some sort of look at that could be useful.

 

[Nym]

The thing if, you can't rely on wheels not slipping and being the size you think they are, you could however just fit 'timing point beacons' and combine this with the speed/position sensors to calibrate while on route. These beacons would simply sit there at whatever intervals, and their position be known to the ERTMS console, so it can calibrate it's position, or if you don't even want to do any calibratring, you could use the beacons to form mock track blocks.

 

[OliverS]

Because of the reception problems of moving vehicles and there can be stretches of track with little cells or non 3G cells. If your remotely monitoring the trains CCTV you dont want it to keep cutting out. Anyway all they have to do is allocate more radio spectrum and a backwards compatible update to the standards such that basic services would be in the existing allocation while additional services would be in the new spectrum.

Surely they could just put in smaller cells? If the problem is around hubs then add lots of pico cells (or possibly micro cells) and lower the power and you get more bandwidth. More cost of course as you need more base stations but that is what the phone networks do.

 

[68000]

Yes but public networks are revenue earning, this is not revenue earning. Also public networks do not care which cell you camp onto, GSM-R does.

 

[DownSouth]

Indeed, the original ERTMS steering commitee came to the conclusion that it was not viable at the time, however the Americans have done a great deal of very interesting work on differential GPS in connection with Positive Train Control, so some sort of look at that could be useful.

It could also be worth considering Inertial Navigation System (INS) which is used on aircraft and submarines. No need for satellites means perfect operation in tunnels, you just need precise coordinates at depots for the system to be calibrated once every few weeks.

 

[HSTEd]

It could also be worth considering Inertial Navigation System (INS) which is used on aircraft and submarines. No need for satellites means perfect operation in tunnels, you just need precise coordinates at depots for the system to be calibrated once every few weeks.

Actually its not practical for more than a few minutes at a time on railways, because submarines only have to worry about not blundering into the shore or a sea mount, and can be quite conservative about how much they "trust" the system. If an SSBN has to fire its missiles from underwater (as it does as a matter of course) it can either stick its periscope up any time over the proceeding half hour or so to reset its INS using a satellite fix or it can even fire "half blind".

If this is a last resort counter-value strike (implied by the lack of GPS coverage) whether the missile misses its aim point over the centre of a city by 100m or by 200m makes little difference.

Aircraft don't need to use INS to perform any precision maneuvres, being able to use GPS for this functionality.

Unfortunately if the INS drifts out by 5 metres or even less its useless as it would now be showing the train as being on the wrong line.

 

[OliverS]

Yes but public networks are revenue earning, this is not revenue earning. Also public networks do not care which cell you camp onto, GSM-R does.

Bandwidth costs money. You can either pay for more spectrum, or you can use more, lower powered, cells. Spectrum isn't free, it is a very limited resource and therefore valuable.

And public networks do care which cell you camp onto, there are algorithms to work out the best cell for you to be in and to shift you to a super-cell if necessary. But that is all part of GSM (I believe) so is inherited by GSM-R.

 

[Hydro]

We already use a combination of inertial and GPS location monitoring for the test trains. It works well enough for our purposes, but does require continuous monitoring at the moment due to hangups on junctions, and occasional IMU (Inertial Measurement Unit)hiccup that indicates the train is on the wrong line. These must be manually overridden. As time goes on, the software is refined through reporting of these overrides and slowly they get worked out of the system.

For the system to work properly, both GPS and inertial location systems must be working correctly. If one goes down, it tends to throw the whole system out.

For the most part, it works, and it works very well. It's very accurate. It will nail the trains position on the correct line, at the correct mileage. We overlay these details onto a rail network map of the UK as a navigation aid.

 

[68000]

Bandwidth costs money. You can either pay for more spectrum, or you can use more, lower powered, cells. Spectrum isn't free, it is a very limited resource and therefore valuable.

And public networks do care which cell you camp onto, there are algorithms to work out the best cell for you to be in and to shift you to a super-cell if necessary. But that is all part of GSM (I believe) so is inherited by GSM-R.

I am sorry, from a revenue earning point of view, the public networks do not give a **** what cell you are on. As long as you can make a call, you will be billed. GSM-R uses the same algorithms and more as they need the mobile to be on specific cells to make the correct call routing therefore GSM-R is much more specific about what cells the mobile camps onto.

 

[OliverS]

I am sorry, from a revenue earning point of view, the public networks do not give a **** what cell you are on. As long as you can make a call, you will be billed. GSM-R uses the same algorithms and more as they need the mobile to be on specific cells to make the correct call routing therefore GSM-R is much more specific about what cells the mobile camps onto.

Public networks care a lot, as they have to make sure that you can make a call in order to bill you. Yes GSM-R cares more, but to say that GSM does not care is a way to lose money as a network.

However the point still stands, if you want more bandwidth you can either add spectrum or cells. Both cost money, adding spectrum involves consuming a finite resource and is therefore in the long term not as scalable. But adding cells has more infrastructure cost and might be more cost-effective as a short term fix.