Should FOCs be required to fit Electronically controlled pneumatic (ECP) brakes?

[MarkyT]

In order for train B to be safely spaced behind train A, a notional control system has to know where the rear end of A is. With CBTC, as used routinely on metros, standard length trains are used that are 'self aware' of how long they are. Thus the control system can easily extrapolate a rear position from the front cab position, derived relative to trackside digital beacons using odometry or similar methods. ECP may be able to determine better real-time train continuity as well as improve braking performance, but it can't output a reliable SIL4 equivalent train length determination of variable length freights accurate enough for train spacing or junction locking release purposes. Some fixed train detection is here to stay I'm certain, at least for safe junction locking. Perhaps, when a train is leaving such a junction area, systems could also use that fixed train detection section going clear to take an instantaneous measurement of the length of the train, a value that could then be stored and used throughout the subsequent plain track section for spacing calculations.

 

[HSTEd]

ECP may be able to determine better real-time train continuity as well as improve braking performance, but it can't output a reliable SIL4 equivalent train length determination of variable length freights accurate enough for train spacing or junction locking release purposes.

If the train knows what wagons it is made up of, and each wagon knows how long it is, why can't it?
Apparently each vehicle is required by the specification to store and make available a variety of important information, including its "Reporting Mark", length over stretched couplers, empty and loaded maximum weights, number of axles on the vehicle and various other bits of pieces related to its braking performance.

Some fixed train detection is here to stay I'm certain, at least for safe junction locking. Perhaps, when a train is leaving such a junction area, systems could also use that fixed train detection section going clear to take an instantaneous measurement of the length of the train, a value that could then be stored and used throughout the subsequent plain track section for spacing calculations.

That's why I said Hybrid Level 3.
But using only axle counters near junctions/level crossings still saves an enormous amount.

As I said, no need for any trackside signalling gear for 20km south of the south end of Stoke Tunnel, as an example.

You could probably use an axle counter to confirm how many axles were in a freight train, which the train should know - if the numbers match you know that every vehicle is properly communicating with the cab equipment.

 

[Belperpete]

Perhaps, when a train is leaving such a junction area, systems could also use that fixed train detection section going clear to take an instantaneous measurement of the length of the train, a value that could then be stored and used throughout the subsequent plain track section for spacing calculations.

The trouble is that the trackside knows when the rear of the train clears the axle counter, and that information would need to be transmitted to the train for it to check its length. The transmission delay would cause a significant uncertainty to the calculation. It can take an axle counter about a second or so to report that it is clear, then add on the interlocking processing time and Radio Block transmission times, and you could have a significant amount of potential delay. The faster the train is travelling, the greater the uncertainty that delay would add to the train length calculation. Similar issues if you do it the other way around, with the train reporting to the interlocking when it thinks it should be clear, so the interlocking can double-check with the axle-counter status.

 

[Belperpete]

ECP systems can also perform train completeness checks to the standard required by ETCS for the removal of continuous trackside train detection equipment.

Can it? What is your source for that?

Essentially all modern MUs, which will soon be essentially all MUs on the network, have train completeness check equipment suitable for ETCS L3H deployment.

Agreed that modern MUs do a train completeness check, but I was not aware it was to a SIL level suitable for ETCS. Likewise I know that modern MUs can determine how many units are coupled, to calculate train length, but I was not aware that this was done to an interlocking SIL standard.

 

[JeffH16]

No?
I demand value for money for the taxpayer.

We can either require a handful of freight operators to spend a pittance, or even pay them the pittance, to install modern ECP technology on the relatively small freight fleet.
Or we can continue spending colossal sums of money on enormous train detection infrastructure that only exists for the benefit of a small number of trains

Essentially all modern MUs, which will soon be essentially all MUs on the network, have train completeness check equipment suitable for ETCS L3H deployment.
Soon the only holdouts will be Freight and the Sleepers.

The public currently accepts spending billions a year keeping the system going in its current form, it is not a good idea to assume that this will always be so.
And even if it was always so, it is still immoral to just assume that public money will continue to rain from heaven and there is no duty to ensure that it is not wasted.

Road Hauliers also currently only pay for around 1/3 of the damage they cause to roads, through fuel duty and excise duty. Would you like to see their costs upped so as to demand value for the tax payer as well?

 

[edwin_m]

If the train knows what wagons it is made up of, and each wagon knows how long it is, why can't it?

Of course it can do that in principle with appropriate data protocols. As other posters are pointing out, it may not be able to do so with adequate safety integrity - crudely speaking it's not reliable enough to rely on in a situation where an incorrect reading could be a danger to life and limb.

The task itself isn't onerous and the system could be re-engineered to do that with high integrity digital processors, something like a trackside module for one of the computer-based interlocking systems. But that might make it more expensive to fit, and power-hungry, and someone would have to pay the development costs.

 

[Nym]

And yet these systems are deployed in Australia, South Africa and Canada, amongst others?

All of whom use the North American not BSI and Related standards for railways.

Surely a battery charger's power consumption is at least passingly related to the battery it is charging, and what the projected battery charger availability is?

Hence the term, minimum.

 

[MarkyT]

Of course it can do that in principle with appropriate data protocols. As other posters are pointing out, it may not be able to do so with adequate safety integrity - crudely speaking it's not reliable enough to rely on in a situation where an incorrect reading could be a danger to life and limb.

The task itself isn't onerous and the system could be re-engineered to do that with high integrity digital processors, something like a trackside module for one of the computer-based interlocking systems. But that might make it more expensive to fit, and power-hungry, and someone would have to pay the development costs.

Also a major reliability issue. Each one of those wagons would have to reliably report to the managing computer at a frequency or else a managed failsafe response would be required. Completeness is easy. The whole train is either complete, or the back of the train remains in the last known position. Length is a more difficult problem. How do you determine the coupling order in a radio system where every wagon individually communicates with the head. Maybe a short-range radio link at each wagon end that can only work across a typical coupling distance and forms a chain down the train. There would be some latency issues in that no doubt. Is there a danger a passing wagon on a parallel track might suddenly appear to be in the consist? Maybe a consist could be 'locked' when being formed up so comms only works between the chain of wagons that are set up and coupled (both physically and logically) in the yard.

 

[edwin_m]

Also a major reliability issue. Each one of those wagons would have to reliably report to the managing computer at a frequency or else a managed failsafe response would be required. Completeness is easy. The whole train is either complete, or the back of the train remains in the last known position. Length is a more difficult problem. How do you determine the coupling order in a radio system where every wagon individually communicates with the head. Maybe a short-range radio link at each wagon end that can only work across a typical coupling distance and forms a chain down the train. There would be some latency issues in that no doubt. Is there a danger a passing wagon on a parallel track might suddenly appear to be in the consist? Maybe a consist could be 'locked' when being formed up so comms only works between the chain of wagons that are set up and coupled (both physically and logically) in the yard.

I think we're discussing a cable-based system here, which would address some of those issues and also deliver power down the train but is vulnerable to damage. As I mentioned somewhere above, multiple working of locos distributed down the train can be done by radio (certainly used to be, not sure of ECP has superseded it) but that must have some safeguard such as setting up the head end radio with the numbers of all the locos down the train.

 

[Nym]

I think we're discussing a cable-based system here, which would address some of those issues and also deliver power down the train but is vulnerable to damage. As I mentioned somewhere above, multiple working of locos distributed down the train can be done by radio (certainly used to be, not sure of ECP has superseded it) but that must have some safeguard such as setting up the head end radio with the numbers of all the locos down the train.

They also still have a brake pipe...

 

[MarkyT]

The trouble is that the trackside knows when the rear of the train clears the axle counter, and that information would need to be transmitted to the train for it to check its length. The transmission delay would cause a significant uncertainty to the calculation. It can take an axle counter about a second or so to report that it is clear, then add on the interlocking processing time and Radio Block transmission times, and you could have a significant amount of potential delay. The faster the train is travelling, the greater the uncertainty that delay would add to the train length calculation. Similar issues if you do it the other way around, with the train reporting to the interlocking when it thinks it should be clear, so the interlocking can double-check with the axle-counter status.

Many latency issues I agree which is why I'm a fan of fixed train detection techniques on main line railways. Fixed blocks are generally fine for capacity in these applications where platform occupancy always remains the dominant constraint. I'd propose a 'modern permissive' overlay in such platform areas where forward-facing radar/lidar supplements 'drive on sight' for entering a platform while the previous train is still departing. Elsewhere the industry could concentrate on making axle counter technology cheaper. Axle counter sensors could use wireless links, and movement authority interlocking management and radio equipment could also become more decentralised and distributed to the trackside using distributed processing and '5G style' comms techniques, a bit like old-style relay auto sections in fact, to reduce latency and increase scalability. By contrast, centrally mediated moving block movement authorities seem a computing overhead and communications latency nightmare. Instead of trying to get rid of most trackside signalling infrastructure (impossible because there will always have to be balises and radio systems), let's try and make it smaller, cheaper, lower-powered, more standardised, integrated, distributed and intelligent. Have I got enough buzzwords in there yet?

 

[Irascible]

Trains are at least inherently predictable - known path & acceleration doesn't change much ( compared to say, a car or a combat aircraft, two other areas where much work on path prediction is done ), so unless you're talking latencies of multiple seconds I don't see that being a problem - if your network has latencies of that order then you've got a bigger problem already. Distributed systems can be more redundant but also have many more points of failure - is that better than having a centralised system with a basic fallover? I really am interested in alternative architecture ideas for this problem.

Fitting every non-MU vehicle with a super-cheap GPS receiver that tells central control what it is and it's position regularily would mean a train of random vehicles wouldn't need to discover it's own length. Pretty much every cellphone has the tech you'd need to do it, economies of scale are already there. You could crosscheck with nearby vehicles for some sanity checking rather than relying on a single data source ( the train head ) too. I'm sure there's 10s more ways of solving the train length problem.

 

[edwin_m]

They also still have a brake pipe...

The brake pipe, assuming it's still configured like a traditional train pipe, provides a backup if the ECP fails "wrong side" - though it probably won't give the same stopping distance so drivers shouldn't rely on the ECP in situations such as approaching a red signal. But the pipe is of no use in reporting the train length and if the ECP modules on the wagons are anything less than Safety Integrity Level 4 then there is a chance the wrong length will be reported. I assume also the ECP cable at the end of the train has to be plugged in somewhere not left hanging, so the system can distinguish a genuine end of train from a divided train.

In relation to radio control of locos from the head end the brake pipe also provides some protection, as I assume there is still a pressure switch to cut power if there is no brake pressure so any loco "going rogue" would be stopped by an emergency brake application. However the same system could also be used for controlling a consist from a handheld unit on the ground, in which case no such fallback is possible.

 

[HSTEd]

The brake pipe, assuming it's still configured like a traditional train pipe, provides a backup if the ECP fails "wrong side" - though it probably won't give the same stopping distance so drivers shouldn't rely on the ECP in situations such as approaching a red signal. But the pipe is of no use in reporting the train length and if the ECP modules on the wagons are anything less than Safety Integrity Level 4 then there is a chance the wrong length will be reported. I assume also the ECP cable at the end of the train has to be plugged in somewhere not left hanging, so the system can distinguish a genuine end of train from a divided train.

The train determines that a divided train has occured because the wagon that should be last in the chain has gone "missing".
Whilst intermediate wagons only report relatively irregularly, unless they have a fault, the designated "end of train" wagon is required to report status on the bus, every second.

Also remember if an ECP controller fails wrong side, that only effects that specific wagon - so if one wagon fails wrong side and nine fail right side (as an example), braking performance will still be quite good.

I think we're discussing a cable-based system here, which would address some of those issues and also deliver power down the train but is vulnerable to damage. As I mentioned somewhere above, multiple working of locos distributed down the train can be done by radio (certainly used to be, not sure of ECP has superseded it) but that must have some safeguard such as setting up the head end radio with the numbers of all the locos down the train.

As far as I am aware, every company that uses ECP has moved to it for distributed power control as well.
Supposedly it has cut the time requierd to establish distributed power control for the train rather drastically.

And it doesn't need much extra equipment compared to the radio gear.

You already have the train line after all! All it takes is a box that an MU control cable plugs into.

 

[Romsey]

Having just browsed this thread, has it got lost from a technical geeks forum?
You're really overthinking freight operations.

1 All this fancy electronic would need maintaining. Wagons are not cossetted, they get quick exams and tests and an annual brake exam and overhaul.
2 Two line AB ( Brake and Main Res) would improve braking performance. ( Although it was stripped off earlier AB freight vehicles in the UK to save maintenance costs.)
3 I'm not certain how or why wagons would need to be part of the ECTS equipment. That's on the loco and clearing track circuits / axle counter sections provides the clearance for the interlocking to clear and permit the next route to be set.
4 If you're really fussed about knowing where the rear of the train is there is always the concept of a "smart fred"
( FRED? - flashing rear end device) which provides radio feedback to loco controls and can apply brakes from the rear of the train as well as the loco at the front. FREDS have largely been replaced in North America by distributed power with mid and rear locos controlled by radio from the leading cab.
5) If a wagons breaks away from the rear of the train, the brake pipe is parted and both portions of the train stop. That is a basic going back to the earliest days of continuous brakes in the 1870's.

Think Meerkat - "Simples"

 

[HSTEd]

3 I'm not certain how or why wagons would need to be part of the ECTS equipment. That's on the loco and clearing track circuits / axle counter sections provides the clearance for the interlocking to clear and permit the next route to be set.

Because it would be nice to dispose of the enormously complex and expensive trackside train detection infrastructure?
WHich pretty soon, if not already in many places, will exist pretty much solely for freight operations.

 

[Nym]

Perhaps one should have considered non passenger vehicles when conceiving this ETCS plan for the UK...
Oh, hang on, wrong country for that

 

[Romsey]

Because it would be nice to dispose of the enormously complex and expensive trackside train detection infrastructure?
WHich pretty soon, if not already in many places, will exist pretty much solely for freight operations.

OK, it's expensive but the detection system is proven to be safe. Remember that conventional detection (track circuits and axle counters) systems are used on HS1 and for fringe areas with AB / TCB on the Cambrian.
Track circuits don't check train length, they are either occupied or clear. Axle counters count the axles in and out of a section to create occupied or clear for a section. On or off, 1 or 0 is what is needed for a logic system whether it is a computer of a relay interlocking.

Any just how much would the programming for each interlocking costs to include train lengths in the logic, then the radio system enhancements on GSM-R and all the kit on every vehicle?

ECTS works in conjunction with the leading cab of any train formation, not every vehicle. Due to the costs, I doubt that we will see anything beyond ECTS 2 on densely trafficked lines south of the Thames in the 20 years.

Another poster suggested driving on line of sight in some areas?
No, just NO. That is at complete variance with all the development of signalling going back to the first interlocking frames of around 1860. It's noticeable that various tram systems now have some form of signalling around junctions and inter-worked with road signals, moving away from their traditional line of sight operations.

There is nothing wrong with suggesting new technology, but it does help to understand how and why what is used now had developed.

 

[HSTEd]

OK, it's expensive but the detection system is proven to be safe. Remember that conventional detection (track circuits and axle counters) systems are used on HS1 and for fringe areas with AB / TCB on the Cambrian.

HS1 is built fundamentally with 90s technology though.
We can do rather better now.

As for fringe areas, the simple answer being dont have fringe areas?

Any just how much would the programming for each interlocking costs to include train lengths in the logic, then the radio system enhancements on GSM-R and all the kit on every vehicle?

Interlockings taking account of train lengths has been done for decades though, just not commoly in heavy rail applications because train completeness checks are a pain.

We arent talking about ETCS gear on every freight wagon, we just need a suitably reliable system for detecting train completeness reliably.
Train length detection is a nice to have, but even if we just assume all freight trains are 800m (or whatever the maximum length is these days) long, we still get major benefits over existing systems.
Since when the only thing you need to separate blocks is a eurobalise fitted to a sleeper, you can put them all over the place like confetti.

ECTS works in conjunction with the leading cab of any train formation, not every vehicle. Due to the costs, I doubt that we will see anything beyond ECTS 2 on densely trafficked lines south of the Thames in the 20 years.

ETCS Hybrid L3 will be enormously cheaper than L2, by dispensing with the huge trackside train detection infrastructure that it requires!
Indeed L2 has a not-great business case because you only save the signal display system, and still need a huge trackside infrastructure, just with less things plugged into it.

No, just NO. That is at complete variance with all the development of signalling going back to the first interlocking frames of around 1860. It's noticeable that various tram systems now have some form of signalling around junctions and inter-worked with road signals, moving away from their traditional line of sight operations.

And yet Manchester Metrolink functionally abandoned their entire train signalling system and reverted pretty much entirely to line on sight driving on plain line?
Putting axle counters on junctions and at level crossings (where signalling gear must be provided regardless) still cuts the equipment requirements rather enormously.

There is nothing wrong with suggesting new technology, but it does help to understand how and why what is used now had developed.

We have track circuits and axle counters because trains had no reliable way of determining if they are no longer complete and informing the signalling system of this fact.
We now have the capability to do that, so the original rationale starts to crumble.

 

[Nym]

HS1 is built fundamentally with 90s technology though.
We can do rather better now.

As for fringe areas, the simple answer being dont have fringe areas?

So convert the whole network at once overnight / christmas?

 

[MarkyT]

Because it would be nice to dispose of the enormously complex and expensive trackside train detection infrastructure?
WHich pretty soon, if not already in many places, will exist pretty much solely for freight operations.

Nowhere on main line infrastructure in the UK, and possibly anywhere else does balise based train detection apply today. There is research ongoing for a hybrid system that could plausibly replace some sections with virtual fixed sections for a limited number of trains that can do it today, and while the tech may exist as a shadow overlay somewhere it is definitely not a proven system in passenger service.

OK, it's expensive but the detection system is proven to be safe. Remember that conventional detection (track circuits and axle counters) systems are used on HS1 and for fringe areas with AB / TCB on the Cambrian.
Track circuits don't check train length, they are either occupied or clear. Axle counters count the axles in and out of a section to create occupied or clear for a section. On or off, 1 or 0 is what is needed for a logic system whether it is a computer of a relay interlocking.

Any just how much would the programming for each interlocking costs to include train lengths in the logic, then the radio system enhancements on GSM-R and all the kit on every vehicle?

ECTS works in conjunction with the leading cab of any train formation, not every vehicle. Due to the costs, I doubt that we will see anything beyond ECTS 2 on densely trafficked lines south of the Thames in the 20 years.

Well put. My position is that I think the virtual fixed block paradigm has some merit for very high capacity scenarios like Thameslink where some complexity of very short sections could be avoided and the vast majority of trains are uniform passenger units, but I think the best bang for buck might be had by concentrating on developing cheap, high reliability, standardised modular axle-counter based fixed block systems.

Another poster suggested driving on line of sight in some areas? No, just NO. That is at complete variance with all the development of signalling going back to the first interlocking frames of around 1860. It's noticeable that various tram systems now have some form of signalling around junctions and inter-worked with road signals, moving away from their traditional line of sight operations. There is nothing wrong with suggesting new technology, but it does help to understand how and why what is used now had developed.

That was me, but please understand I'm not talking about it generally at all, and especially not through junctions, I propose it as just a special system for entering platforms from the run-in end if still partly occupied by the train in front, and it wouldn't be based on sight alone at all. Think of it as a special limited speed permissive mode overlay, enhanced with some kind of forward based scanning sensor system such as radar or lidar, that could avoid the multiple fixed block division of platforms as implemented on Thameslink. Otherwise I'm a fixed 'absolute' block man all the way and a signal engineer of over 30 years experience!

 

[MarkyT]

ETCS Hybrid L3 will be enormously cheaper than L2, by dispensing with the huge trackside train detection infrastructure that it requires!
Indeed L2 has a not-great business case because you only save the signal display system, and still need a huge trackside infrastructure, just with less things plugged into it.

A notional L3 system will still need balises and a very comprehensive data radio system for full reliable coverage. It will also almost certainly have a plethora of fixed stop markers because movement authorities need limits to be manageable, especially in failure scenarios. A simple, low power axle counter sensor nodes for each fixed block boundary could be very easily 'plugged in' (wirelessly if neccessary) to that network and may actually represent better value, being simpler and more reliable than a great deal of extra complexity added to every train.

 

[HSTEd]

A notional L3 system will still need balises and a very comprehensive data radio system for full reliable coverage.

But the balises required for such a system are entirely passive devices.
They aren't really comparable to the axle counters and other equipment that must have redundant power supplies, even if they were to use GSM-R to communicate to the block control post.

Also for ETCS Level 2 to work we need a very comprehensive data radio system for full reliable coverage......

A simple, low power axle counter sensor nodes for each fixed block boundary could be very easily 'plugged in' (wirelessly if neccessary) to that network and may actually represent better value, being simpler and more reliable than a great deal of extra complexity added to every train.

A "great deal" of complexity added to every freight train - which represent a fraction of the trains on the network. Only 600 freight trains run per day after all.
IEPs have the train completeness functionality built in, as will most, or all, other modern multiple units.

How many thousand axle counters will you need?

Well put. My position is that I think the virtual fixed block paradigm has some merit for very high capacity scenarios like Thameslink where some complexity of very short sections could be avoided and the vast majority of trains are uniform passenger units, but I think the best bang for buck might be had by concentrating on developing cheap, high reliability, standardised modular axle-counter based fixed block systems.

That description increasingly describes very large portions of the British railway network though!

Privatisation saw an increase in rolling stock diversity, but that is rapidly diminishing again, and the performance distinctions between "suburban" or "local" trains and "intercity" ones is vanishing as everyone ends up against the hard limits of power availability and tractive effort limitations. In the BR era would people seriously think all new diesel multiple units would be rated 100mph?

More intensive operation of existing infrastructure has been a watchword since at least the 80s, and I don't think its going to stop.
Building new railways is extraordinarily expensive, fitting balises every 100m (or indeed less!) is not.

 

[MarkyT]

A "great deal" of complexity added to every freight train - which represent a fraction of the trains on the network. Only 600 freight trains run per day after all.
IEPs have the train completeness functionality built in, as will most, or all, other modern multiple units.

No mainline train on the network, even those that already have ETCS fully enabled on board are currently actually ENABLED and certified to report their length for block clearance purposes, not least because there is no infrastructure that requires this with which they have been tested. That specific L3 capability will have to be proven exhaustively before being introduced. Just because ETCS has the conceptual protocols for reporting this data across the track-train boundary in a standard way does not mean the functionality is actually already built into control systems either side of that boundary, or that it is practical to do that very quickly. Network Rail is wholly set up as a fixed train detection railway today and even if a policy change was made tomorrow it would take decades to change the installed base, and will require negotiation with a host of operating companies and maintenance organisations that have rolling stock and moving plant that can't easily support this feature. We've been long enough getting to cab based already. It's not time to introduce yet another variable into the mix. Signalling schemes are being installed currently that have expected lifespans of up to 40 years. I think you are being wildly optimistic about the practicality of what you suggest.

 

[LAX54]

And yet these systems are deployed in Australia, South Africa and Canada, amongst others?

Just because they use those in those Countries, does not mean they are better than ours, the UK does not have the reputation of the safest railway in the world for no reason.

 

[Belperpete]

Train length detection is a nice to have, but even if we just assume all freight trains are 800m (or whatever the maximum length is these days) long, we still get major benefits over existing systems.

That is not an unreasonable suggestion. For any train that cannot determine its length with suitable integrity, the system just assumes a set maximum train length. Assuming that only a small proportion of the trains are doing this, it shouldn't have a significant impact on headways - at least, hopefully not enough to counter-balance the other improvements. That just leaves train integrity, which is relatively easy to do with an end-of-train device.

The only problem I foresee would be at junctions. A say 400m long train could be standing clear of a diverging junction in rear, but if the system is assuming it is 800m long it will think the rear of the train is still standing over the junction, and blocking other moves. However, initially at least, I think that most points and junctions are likely to be protected by axle-counters. Not for safety reasons, but for operational reasons. Axle-counters can locate train position much more accurately, allowing a junction to be released when an on-board system with less accuracy could say the train is still blocking the junction, causing significant delays to other trains.

Of course, track-circuits and axle-counters don't detect the end of train either. They detect the last wheel, but that is not the same as the end of the train. Which is why axle-counter heads aren't positioned at the fouling point of a junction, but at the clearance point. This allows for the worst-case overhang beyond the last wheel.

 

[Belperpete]

Perhaps one should have considered non passenger vehicles when conceiving this ETCS plan for the UK...
Oh, hang on, wrong country for that

ETCS wasn't conceived for the UK - the clue is in the name "European Train Control System".

ETCS levels 1 and 2 are already being successfully rolled out throughout Europe (and that includes the UK). Levels 1 and 2 only detect the front of the train, and so still rely on conventional trackside train detection to detect the rest of the train. Level 3 does away with the conventional (and costly) train detection, and so the train needs to know where both its front and rear are. The ETCS spec leaves it to individual railway companies as to how they achieve that. Level 3 is starting to be rolled out, but mainly on railways with fixed train sets. How to detect where the rear of the train is on railways with variable formation trains is a problem that not only the UK is trying to solve.

 

[Belperpete]

Axle counter sensors could use wireless links

I believe that such systems are already being developed. However, you would still require equipment on the ground, which would need a power supply.

 

[edwin_m]

The train determines that a divided train has occured because the wagon that should be last in the chain has gone "missing".
Whilst intermediate wagons only report relatively irregularly, unless they have a fault, the designated "end of train" wagon is required to report status on the bus, every second.

Also remember if an ECP controller fails wrong side, that only effects that specific wagon - so if one wagon fails wrong side and nine fail right side (as an example), braking performance will still be quite good.

For someone who claims technical background in things such as nuclear technology, your understanding of integrity and RAM seems somewhat shaky. Not to mention the compatibility issues that beset a system like a railway, whereas a power station is never going to be coupled to another power station so they don't need to work together.

Wrong side failure of the ECP controller could result in it failing to apply the brakes, which is a big problem if it's the last wagon in the train, as a broken coupling could result in it running loose. The obvious answer to that is to make the air pipe operate as a conventional air brake as a backup, which also gives compatibility with conventional wagons. But you seem to be opposed to that for some reason.

In respect of train complete, there is another failure mode where an intermediate wagon could decide incorrectly it was the last wagon and transmit accordingly. To be certain that won't happen you need a SIL 4 system.

We arent talking about ETCS gear on every freight wagon, we just need a suitably reliable system for detecting train completeness reliably.
Train length detection is a nice to have, but even if we just assume all freight trains are 800m (or whatever the maximum length is these days) long, we still get major benefits over existing systems.

There are occasions when trains are longer than 800m, for example double headed double trains have been run during diversions to save track capacity. Assuming each freight train is that long would eat up capacity behind it.

I thought we came into this from the premise that we needed ECP to prove train length? If checking the pressure in the brake pipe isn't good enough for train completeness then just jumper two wires down the train and connect them together only at the far end. This could be made so it only worked by plugging into a special socket on the tail lamp. As the original reason for ECP to manage stopping in very long trains doesn't really apply in Europe, don't we just have a solution in need of a problem?

 

[HSTEd]

Wrong side failure of the ECP controller could result in it failing to apply the brakes, which is a big problem if it's the last wagon in the train, as a broken coupling could result in it running loose. The obvious answer to that is to make the air pipe operate as a conventional air brake as a backup, which also gives compatibility with conventional wagons. But you seem to be opposed to that for some reason.

This only protects against a fault when the wagon control system fails, and yet the wagon control system still responds to a fall in air pressure in the train line?
I am skeptical you get a meaningful increase in safety from such a measure, but ok.

And compatability with conventional wagons is not really a positive in this context, and may well be a negative.

In respect of train complete, there is another failure mode where an intermediate wagon could decide incorrectly it was the last wagon and transmit accordingly. To be certain that won't happen you need a SIL 4 system.

Hopefully the people composing the train should know how long the train they are supposed to be driving today is.

And as I suggested above, the first time the train drives over an axle counter leaving the yard, it should be possible to tell the train how many axles it has.
It should know how many axles it has, if those numbers don't match up you have an unknown vehicle or vehicles in the formation.

I thought we came into this from the premise that we needed ECP to prove train length? If checking the pressure in the brake pipe isn't good enough for train completeness then just jumper two wires down the train and connect them together only at the far end. This could be made so it only worked by plugging into a special socket on the tail lamp. As the original reason for ECP to manage stopping in very long trains doesn't really apply in Europe, don't we just have a solution in need of a problem?

So you want to install jumper cables down the entire length of the entire freight fleet?
That will cost a substantial fraction of the cost of the ECP system!