ETCS Roll-Out on ECML South Confirmed

[Agent_Squash]

Depends on what you mean by digital.

A electromagnetic relay is a digital device. A conventional relay has two states (* some specialist relays have three states, but they are still digital): energised and de-energised. It’s a binary digital device. Hence a route relay interlocking is a digital system.

If by digital, you mean computer controlled, as in computer based (that is there is still a human signaller that inputs each route), that describes the British Rail Solid State Interlocking (SSI) system (first full installation was at Leamington Spa in 1985). See https://en.wikipedia.org/wiki/Solid_State_Interlocking.

BR then introduced Integrated Electronic Control Centre (IECC) systems (see https://en.wikipedia.org/wiki/Integrated_Electronic_Control_Centre).

If you mean a computer based signalling system which has an automatic route setting (ARS) system, yet again, this is nothing new. That technology has been in use in this country for many years already.

In terms of train protection systems, we already have the Great Western Automatic Train Protection (ATP) system which uses digital computers and digital data transmissions. And we already have an ERTMS system. See https://en.wikipedia.org/wiki/Europ...ystem_in_Great_Britain#Current_implementation

Digital describes a technology that generates, stores, and processes signals or data in terms of two states: typically represented by values of a physical quantity such as voltage (two different voltages or polarities such as positive and non-positive), magnetic polarization or physical position/state. One state is typically expressed or represented by the number 1, or logic high and the other by the number 0 or logic low. Thus, signals or data transmitted or stored with digital technology is normally expressed as a string of 0's and 1's.

Note that a digital signal does not have to have only two states, but it has to have a limited, fixed and defined number of states.

I am fully aware of what digital means - but thanks for the explanation. Would be good to point out that digital isn’t actually a technology - it is just a reference as to how data is stored. You can store data in a ‘digital’ format without involving a computer at all!

My point still stands though - your original post says that we don’t do cutting edge signalling technology. Crossrail was arguably one of the most complex applications of modern signalling technology, with integration between three signalling systems (more than anywhere else), and it was so advanced as a network to the extent that Mark Wild said it may’ve gone too far.

 

[Sonik]

"Electronically controlled pneumatic" braking, a system with a power and data base running the length of the train that allows independent electronic control of all brakes in the formation, and data passage between vehicles and the locomotives. The air brake pipe is retained solely as a method of continuously refilling the reservoirs in the vehicles. It's been deployed on some heavy haul railways in Australia, South Africa and North America but the up front cost of mass retrofit has held it back elswhere.

With such a system you can provide continuous train completeness checks, transmit multiple working data and any number of other things.

Interesting, thanks.

This does sound like a solution if not a simple one. But likely still cheaper and easier than equipping all freight wagons with transponders or using a dedicated trailer.

The problem AFAICS will be getting the whole fleet equipped, with the fragmented ownership and maintenance of wagons. Think NR/GBR will need to take a lead here.

 

[Annetts key]

I am fully aware of what digital means - but thanks for the explanation. Would be good to point out that digital isn’t actually a technology - it is just a reference as to how data is stored. You can store data in a ‘digital’ format without involving a computer at all!

Sorry, it wasn’t my intention to be condescending.

My point still stands though - your original post says that we don’t do cutting edge signalling technology. Crossrail was arguably one of the most complex applications of modern signalling technology, with integration between three signalling systems (more than anywhere else), and it was so advanced as a network to the extent that Mark Wild said it may’ve gone too far.

Yes, the Crossrail project is one of the most complex applications of modern signalling technology. In the past, the Channel Tunnel was another.

But my point is that the so called state of the art / cutting edge signalling technology is normally either just a new installation of existing technology, or a evolution of an existing technology.

Signalling systems have to be bullet proof. They have to be reliable and safe. They must be absolutely totally fail safe. All this means that very extensive development and testing is needed. Hence the amount of time and money needed to develop new systems from scratch makes this approach unrealistic for most schemes.

In some schemes where the overall system is made up of multiple different systems, the complexity is in interfacing the different systems together so that they work as intended.

When a new or altered signalling system is due to be commissioned, extensive signalling principles testing is required. Some of this can be done by the use of simulation. Or by using a test set-up away from the real railway and providing the system with artificial inputs while monitoring the outputs. But if there are multiple different systems that interface to each other, simulation may be impractical, so some of this testing would have to be carried out on the actual real railway system before it opens to passenger traffic.

In any event, all the equipment that is being commissioned needs functional testing to ensure there are no crossed wires, mixed up cables, incorrect equipment unique addresses etc. etc.

If you are introducing new trains as well, that just adds even more testing, as then you have to ensure that the trains don’t generate interference that can affect the signalling system.

If there are automatic train protection systems or train control systems, that all has to be tested as well.

 

[Jonny]

Something I've heard about ETCS is "Baseline 3" is the most recent version and it requires the use of on-train Doppler for speed detection?

 

[najaB]

Obsolescence is a big problem in railway signalling. The thing is the newer stuff becomes obsolete quicker. It's relatively easy to get a 1930/40s relay serviced but a 10-15 year old circuit board is a different story. Companies now want to sell you the latest new piece of equipment. Not to mention software upgrades which cost a fortune. Things need to modernise but it can end up costing a lot of money.

In theory, at least, modern signalling uses software-defined interlocking - which means that everything runs on general-purpose computers rather than needing application-specific ICs. So replacing equipment is comparatively trivial, one x86 processor is just as good as any other. And at a push you can always virtualise the hardware to completely remove any requirement for specialist hardware.

 

[HSTEd]

In theory, at least, modern signalling uses software-defined interlocking - which means that everything runs on general-purpose computers rather than needing application-specific ICs. So replacing equipment is comparatively trivial, one x86 processor is just as good as any other. And at a push you can always virtualise the hardware to completely remove any requirement for specialist hardware.

It's a matter of time before someone writes an interlocking on Java......

 

[najaB]

It's a matter of time before someone writes an interlocking on Java...

Was that a challenge? Sounds like a job for Real time Java to me.

 

[HSTEd]

Was that a challenge? Sounds like a job for Real time Java to me.

Sometimes the future is truly terrifying to behold, isn't it?

 

[Roast Veg]

That's one way to ensure your software will be out of date before it's installed.

 

[sor]

Was that a challenge? Sounds like a job for Real time Java to me.

given the obsession with inserting (normally public) cloud services into things where it might not make sense, I look forward to each interlocking being an AWS instance away (!)

 

[najaB]

given the obsession with inserting (normally public) cloud services into things where it might not make sense, I look forward to each interlocking being an AWS instance away (!)

That's a ROC, no?

 

[MarkyT]

given the obsession with inserting (normally public) cloud services into things where it might not make sense, I look forward to each interlocking being an AWS instance away (!)

I look forward to reviewing your detailed safety case demonstrating SIL4 for that platform! (the highest safety integrity level, used for highly safety critical systems in railway signalling, medical devices, weapons control systems etc.)

 

[najaB]

I look forward to reviewing your detailed safety case demonstrating SIL4 for that platform! (the highest safety integrity level, used for highly safety critical systems in railway signalling, medical devices, weapons control systems etc.)

I could provide you a safety case, or I could just point out "75% cost savings".

 

[Class 170101]

How will the interfaces between the new signalling on the ECML and the routes off it where existing signalling will remain be managed? For example Peterborough PSB control area to Kings Dyke (a manual signal box)?

I notice that Crossrail tunnel signalling may be having problems with the interface with the GEML and thats only late 1980s / early 1990s IECC style signalling.

 

[Bald Rick]

How will the interfaces between the new signalling on the ECML and the routes off it where existing signalling will remain be managed? For example Peterborough PSB control area to Kings Dyke (a manual signal box)?

I notice that Crossrail tunnel signalling may be having problems with the interface with the GEML and thats only late 1980s / early 1990s IECC style signalling.

presumably the same as the interface between the ETCS on the Cambrian and Conventional signalling at Sutton Bridge Junction. Or the already commissioned interface on the ECML between ETCS and conventional signalling at Finsbury Park.

the problems between the Crossrail CBTC and conventional signalling at Stratford (now resolved AIUI) are more than likely to have been avoided if the Crossrail core had been signalled with ETCS…

 

[MarkyT]

How will the interfaces between the new signalling on the ECML and the routes off it where existing signalling will remain be managed? For example Peterborough PSB control area to Kings Dyke (a manual signal box)?

presumably the same as the interface between the ETCS on the Cambrian and Conventional signalling at Sutton Bridge Junction. Or the already commissioned interface on the ECML between ETCS and conventional signalling at Finsbury Park.

The Peterborough station area and north thereof to Stoke will retain conventional signals with both TPWS/AWS and an ETCS overlay, for handling non-ETCS equipped trains on Nottingham and Leicester lines. Fitted and non-fitted trains will thus be able to interwork with each other. Southbound LNER Azumas will be able to transition to ETCS just south of Grantham and stay in that mode for the transition to markerboard only signalling just south of Peterborough.

 

[Annetts key]

In theory, at least, modern signalling uses software-defined interlocking - which means that everything runs on general-purpose computers rather than needing application-specific ICs. So replacing equipment is comparatively trivial, one x86 processor is just as good as any other. And at a push you can always virtualise the hardware to completely remove any requirement for specialist hardware.

You still need specialist hardware to drive the actual signals with 110V AC, the points machines with 110V/120V/140V DC, etc.
And data links and other communications have to respond in the defined time frame (although as modern systems are supposed to be faster, this should not be a problem, as long as the virtual system is not made out of bloated poorly written code).

 

[najaB]

You still need specialist hardware to drive the actual signals with 110V AC, the points machines with 110V/120V/140V DC, etc.

Specialist in that it's not commonly used in other industries true, but a points machine is a points machine and a relay is a relay - there's not nearly the same degree of manufacturer lock-in as you have with solid-state interlockings and legacy signalling equipment.

 

[Jonny]

It's a matter of time before someone writes an interlocking on Java......

Was that a challenge? Sounds like a job for Real time Java to me.

Java also happens to be a proprietary language.

 

[HSTEd]

You still need specialist hardware to drive the actual signals with 110V AC, the points machines with 110V/120V/140V DC, etc.

Whilst that is non-standard equipment it is also equipment that can be engineered out of commodity components and I doubt that will change in the future.

The challenge is to develop standards for interfaces that can be implemented with components cheap and common enough that they are unlikely to ever go out of production completely.
This is one of the reasons RS-232 or PS/2 still hangs on decades after they became technically "obsolescent" - they're really easy to implement.

 

[najaB]

Java also happens to be a proprietary language.

I thought it had been open-sourced now?

 

[Mordac]

I thought it had been open-sourced now?

Been open source for ages...

 

[Annetts key]

Specialist in that it's not commonly used in other industries true, but a points machine is a points machine and a relay is a relay - there's not nearly the same degree of manufacturer lock-in as you have with solid-state interlockings and legacy signalling equipment.

Whilst that is non-standard equipment it is also equipment that can be engineered out of commodity components and I doubt that will change in the future.

The challenge is to develop standards for interfaces that can be implemented with components cheap and common enough that they are unlikely to ever go out of production completely.
This is one of the reasons RS-232 or PS/2 still hangs on decades after they became technically "obsolescent" - they're really easy to implement.

That’s all easy to say, but supplies of semiconductor parts changes every year. I was only yesterday looking at a particular voltage regulator on electronic suppliers websites earlier this week, only to find that the part I was looking for has become obsolete in the past year or so.

COVID19 and the current situation in the world has badly affected various parts of the electronic industry.

If a particular part is no longer manufactured, and there is no direct exactly the same equivalent, then the design may have to be changed. Plus, that then may make all existing designs and systems/modules far more complicated to repair, assuming that repair is still practical.

How much change in circuitry can occur before the redesigned item has to be subjected to the approvals process?

And if the railway does not buy enough, it may not be profitable for the manufacturer to continue to redesign and manufacture said equipment.

Relays are less of a problem, because the BR specifications are available. So replacement relays should be able to be manufactured, even if they are expensive compared to past prices.

But where Network Rail does not own or have access to the designs (including all relevant intellectual property rights) it can get very difficult once the relevant company stops support.

 

[HSTEd]

That’s all easy to say, but supplies of semiconductor parts changes every year. I was only yesterday looking at a particular voltage regulator on electronic suppliers websites earlier this week, only to find that the part I was looking for has become obsolete in the past year or so.

COVID19 and the current situation in the world has badly affected various parts of the electronic industry.

If a particular part is no longer manufactured, and there is no direct exactly the same equivalent, then the design may have to be changed. Plus, that then may make all existing designs and systems/modules far more complicated to repair, assuming that repair is still practical.

How much change in circuitry can occur before the redesigned item has to be subjected to the approvals process?

And if the railway does not buy enough, it may not be profitable for the manufacturer to continue to redesign and manufacture said equipment.

Relays are less of a problem, because the BR specifications are available. So replacement relays should be able to be manufactured, even if they are expensive compared to past prices.

But where Network Rail does not own or have access to the designs (including all relevant intellectual property rights) it can get very difficult once the relevant company stops support.

Ultimately you have to think less in terms of repairing individual pieces of electronic equipment.
That's not what I'm suggesting.

If an interlocking computer suffers a fault, it matters less that the computer is repairable than that the computer can be replaced with a more modern unit and still be able to talk to adjacent equipment.

RS-232 or CAN bus, as examples, are comparatively simple to the point that they can be implemented with commodity components that are unlikely to ever become truly unavailable, even if the specific components change over time.
It doesn't matter precisely what a CAN-bus controller looks like internally so long as it can comply with the specification for the electrical properties of the bus itself.

Fibre optic data connects are particularly good for this because the only components that need to be available to build such a system are the laser diode and photo-receptor, and they don't even need to be of a specific design beyond being built for a specific wavelength of light - all other details of how precisely the system is engineered are inconsequential.

It is not necessary for the railway to develop and procure large numbers of every individual signalling component, you must only procure adaptor cards to enable conversion between the standard data interchange format and whatever new internal computer standards are popular at the time of procurement. Just as interface cards for RS-232 using PCIe exist now. That is a much less of an economic challenge.

 

[Annetts key]

RS232, RS485, and ethernet are indeed not likely to be problem areas anytime soon. These are used in many industries.

My point is that in the real world outside the virtual world of an interlocking system running on generic computer hardware, the system has to talk to the equipment in the lineside equipment cupboards. A lot of the resignalling of the GWML (and elsewhere) has used solid state interlocking (SSI) data link modules (DLM) along with the relevant lineside copper cables and track side functional modules (TFM) such as signal modules and points modules.

The SSI data link is a form of baseband network. The TFM connect the interlocking system via the SSI data link cables and DLM to the “real” lineside signals, AWS, TPWS, ATP, track circuits, axle counters, points machines etc.

So you can’t in practice use adapter cards in location cupboards. It’s not practical.

Because the TFM drive the list of items I gave earlier, they have to confirm to railway standards. For example, there is a maximum difference in allowable voltage levels between the two independent input circuits internal to a TFM for each external input circuit. If the difference is outside the allowable limit, the module shuts down to a fail safe mode. It’s similar with the outputs, if a voltage is detected on an output that is not supposed to be energised, the module will shut down it’s output interface for safety.

Sorting out modern generic PCs and interconnecting them in the signalling centre / signal box is the easy bit!

 

[HSTEd]

The SSI data link is a form of baseband network. The TFM connect the interlocking system via the SSI data link cables and DLM to the “real” lineside signals, AWS, TPWS, ATP, track circuits, axle counters, points machines etc.

So you can’t in practice use adapter cards in location cupboards. It’s not practical.

Because the TFM drive the list of items I gave earlier, they have to confirm to railway standards. For example, there is a maximum difference in allowable voltage levels between the two independent input circuits internal to a TFM for each external input circuit. If the difference is outside the allowable limit, the module shuts down to a fail safe mode. It’s similar with the outputs, if a voltage is detected on an output that is not supposed to be energised, the module will shut down it’s output interface for safety.

Sorting out modern generic PCs and interconnecting them in the signalling centre / signal box is the easy bit!

Are the functions of a trackside cabinet really that different from the sorts of functionality that generic SIL4 PLCs used in industry have?

Proving circuits and detection of erroneous outputs is fairly standard now I thought.

EDIT:

There will be some short term pain in terms of rewiring existing systems to reduce the number of outputs, but ultimately this is the best solution in the long run.
Hopefully ETCS will help by removing a large part of the panoply of trackside equipment.

Ultimately we could hope to be reduced solely to point machines and axle counters.

 

[Annetts key]

Are the functions of a trackside cabinet really that different from the sorts of functionality that generic SIL4 PLCs used in industry have?

Each SSI TFM has two independent computers for safety. Indeed, most of the circuitry is duplicated for this reason. If either computer or it’s associated circuitry diverges from the other one, one or the other will blow a ‘security’ fuse to kill (disconnect the power to) both the computers in the module. If a module shuts down or the security’ fuse gets blown, and it’s a signal module that feeds a lineside signal that has a red aspect, it will go to ‘red retaining’ mode so that the signal (or signals) will stay lit at their most restrictive aspect.

All TFM will go to a safe state if they do not receive the correct communication from the interlocking. For a signal module, again, it will put any signals to their most restrictive aspect.

The designers tried very hard to make the system as fail safe as possible. Remember, the railways of the day had to demonstrate that a computerised system was just as safe as an interlocking system based on fail safe systems built using relays that were also designed to fail safe (including carbon contacts to prevent any possibly of contacts welding themselves together).

As well as traditional SSI schemes, both Smartlock and Westlock use SSI TFMs.

Here’s some photos of TFM:

 

[Trestrol]

My that's an old Signal Model Mod 15 it's done well to last so long.

 

[Annetts key]

My that's an old Signal Model Mod 15 it's done well to last so long.

Stock photos from Wikipedia commons…

The older modules that are supplied as working spares are now mod state 16, software version D1.

 

[Trestrol]

Stock photos from Wikipedia commons…

The older modules that are supplied as working spares are now mod state 16, software version D1.

GEC/ALSTOM are now Mod 17 and are equivalent to a MK3. Solves the TPWS cascade effect.