Following the Bedford incident, should TPWS - or something providing an equivalent level of protection - be fitted to all remaining unfitted signals?

[MarkyT]

Do you have access to a info on the size of a radio infill unit? I can't seem to find a picture of one and they don't appear to be a prominent part of signal company catalogues.

I'm trying to work out of it fills a cabinet or it's the size of a mobile phone!

I also haven't been able to find any illustration of the actual equipment! I expect they'll be fairly compact, with a serial data connection from LEU to a radio transceiver and antenna.

 

[Belperpete]

Given that AWS is a screen indication in modern vehicles I wondered if the AWS visual could flash between horn cancellation and the next bell and even give a warning sound when power is applied in that state?

That would be a considerable distraction in the many, many cases where a driver is legitimately entitled to accelerate after an AWS warning. And would probably lead to it being routinely ignored.

== Doublepost prevention - post automatically merged: 3 Jul 2026 ==

I agree with you about the initial time and costs of R&D, but I think with a bit of clever thinking we could come up with an effective and cheap overlay system using existing equipment that is already PADS approved and configure it to work in a different way. Keep it internal to NR/GBR and roll it out on any project that alters existing layouts in legacy areas. Now the argument that would probably kill the idea is the risk benefit/cost as it always does but I think it's worth exploring at least even if it does turn out to be unfeasible.

If you use something in a way that it wasn't approved to be used for, then it isn't approved! The approval covers not just the bit of equipment, but how it is used.

Don't forget that TPWS was envisaged to be a simple overlay system. A voltage sensor connected in parallel with the red aspect to switch the grids, and fault-monitoring by a volt-free contact connected in series with the signal's existing first-filament proving circuit. All the control equipment mounted in its own standalone case, with just a few wires connected onto existing circuit links in the existing signalling equipment case. And on the trains, the TPWS box was designed to be a bolt-on, plug-coupler-compatible replacement for the traditional AWS box, with the new box designed to do both TPWS and AWS. Just wire up the new TPWS receiver and DMI, independent of everything else. The concept really couldn't have been much simpler and easy to implement. But look how much time and money that took to roll out across the network.

The idea of developing another new system, or another significant upgrade to TPWS and/or AWS, in any less time or money is pie in the sky.

== Doublepost prevention - post automatically merged: 3 Jul 2026 ==

I've had the same thought - fit TSS at signals following junctions where the aspect has been 'artificially' restricted, but where the following signal will in most cases be encountered at Green.

Junction signalling principles have changed a lot over the years. At one time, it was practice to hold the junction signal at yellow. Later practice was to allow the signal to step up to green or double yellow after the train had passed the AWS magnet (and so got a warning). Later still, the principle was changed to allow step up after the point at which the driver should know that he is taking the diverging route. Examples of all these will be found across the network, depending on the principles in place at the time the signalling was commissioned at a given site.

My suggestion would be to update existing installations that hold the junction signal at yellow, to allow them to show the driver the appropriate aspect depending on the signal ahead. Routinely getting a yellow and then a green on the next signal is just asking for trouble, and is ridiculous in situations where the driver can SEE the signal ahead at green when passing the yellow.

== Doublepost prevention - post automatically merged: 3 Jul 2026 ==

Agree with this - there is definitely an argument that all signals immediately following crossovers in this scenario should get TPWS fitted.

It was a pure flook that the first train stopped where it did. Had it stopped one signal section later, for example, I suspect that the collision might still have happened, and so would not have been prevented by TPWS on the first signal after the junction. So do we need the first two signals after? Or three? ......

And why are you limiting it to first signals after a crossover? What about first signals after any converging junction? First signals after a station, or anywhere else a train might stop and then unexpectedly catch up with the train ahead?

We don't yet know enough about this accident to know if TPWS would even have mitigated the consequences, let alone prevented it. All we know so far is that braking began 9 seconds beforehand, and reduced the train speed. We don't know what type of brake applications were made, what initiated it/them, and whereabouts the train was when braking was initiated. Anything else is pure supposition at this stage.

 

[Annetts key]

Don't forget that TPWS was envisaged to be a simple overlay system. A voltage sensor connected in parallel with the red aspect to switch the grids, and fault-monitoring by a volt-free contact connected in series with the signal's existing first-filament proving circuit.

On the TPWS installations fed from relay based interlocking that I have worked on (all Western Region free wired interlocking) bar one (which used the separate TPWS signal box unit), in practice, it's been more complex.

The 'fault-monitoring' contact has been wired into the main lamp proving circuit normally by providing an additional relay, a VCR in the relay room for the interlocking because (1), not all signals have a GECPR (this being the Western Region circuit name) and (2), another contact of the HR and if the signal has a subsidiary position light aspect also a contact of the (S)GR is needed in the VCR circuit so that when the TPWS is not energised, that is the signal is showing a proceed aspect, the TPWS 'fault-monitoring' contact being open does not mess up both the signal in rear and the signallers indication for the signal.

Only buffer stops and separate speed restriction OSS were wired into the first filament failure circuit (on Western this being the G(M)ESR circuit).

Additionally, in some installations extra special controls had to be provided, because of either where the track circuits (IRJs or the turned zone for ASTER U/SF15/TI21) were, where it was possible to actually fit the TPWS TSS loops ('racks') because of either points or an existing ATP signal beacon or other reasons, otherwise, the signal could go back to red as the front of the train passed it, which caused the TPWS to energise, which then tripped the brakes on the train because the receiver under the train was still in range of the track mounted loops ('racks') causing the driver to be a bit confused because they had only seen a proceed aspect...

Indeed, the very first two TPWS signal installations in the area where I was, were taken out of use after being in service for only one day because of problems. It was not until a solution was found and implemented that these two installations were brought back into use.

All the control equipment mounted in its own standalone case, with just a few wires connected onto existing circuit links in the existing signalling equipment case.

This may have been the case for some installations, but it was not the same everywhere. Lots of relay rooms (mostly those that contained the local interlocking) had to have extra relays installed and extra 12 core cables (to the new TPWS cases) and some TPWS modules were fitted either in relay rooms or very occasionally in existing location cupboards.

That's the problem with trying to come up with a simple low cost solution, there isn't a standard colour light signal installation. So you can't have a standard universal new system which wires into the existing signalling system. The nearest that I have seen is how the GWML ATP uses current transformers made to fit on existing 0BA or 2BA link strips. Even that needed one extra circuit for junction signals if one route did not have a junction or route indicator.

The idea of developing another new system, or another significant upgrade to TPWS and/or AWS, in any less time or money is pie in the sky.

Junction signalling principles have changed a lot over the years. At one time, it was practice to hold the junction signal at yellow. Later practice was to allow the signal to step up to green or double yellow after the train had passed the AWS magnet (and so got a warning). Later still, the principle was changed to allow step up after the point at which the driver should know that he is taking the diverging route. Examples of all these will be found across the network, depending on the principles in place at the time the signalling was commissioned at a given site.

My suggestion would be to update existing installations that hold the junction signal at yellow, to allow them to show the driver the appropriate aspect depending on the signal ahead. Routinely getting a yellow and then a green on the next signal is just asking for trouble, and is ridiculous in situations where the driver can SEE the signal ahead at green when passing the yellow.

It was a pure flook that the first train stopped where it did. Had it stopped one signal section later, for example, I suspect that the collision might still have happened, and so would not have been prevented by TPWS on the first signal after the junction. So do we need the first two signals after? Or three? ......

And why are you limiting it to first signals after a crossover? What about first signals after any converging junction? First signals after a station, or anywhere else a train might stop and then unexpectedly catch up with the train ahead?

We don't yet know enough about this accident to know if TPWS would even have mitigated the consequences, let alone prevented it. All we know so far is that braking began 9 seconds beforehand, and reduced the train speed. We don't know what type of brake applications were made, what initiated it/them, and whereabouts the train was when braking was initiated. Anything else is pure supposition at this stage.

Agree fully with everything said in this part that I quoted.

 

[rmHawk765]

And why are you limiting it to first signals after a crossover? What about first signals after any converging junction? First signals after a station, or anywhere else a train might stop and then unexpectedly catch up with the train ahead?

The truth is that any train protection system only makes sense if installed everywhere. A collision can happen literally anywhere on a running line as you say. The whole risk-to-cost ratio nonsense that was probably used to conclude that TPWS is only needed at junction signals and select high-risk areas needs to be abandoned completely. If the risk exists and can be mitigated by a train protection system it should be mitigated, no matter whether trains pass through there once or one thousand times a day. I'm almost certain France and Germany didn't base their nationwide KVB/PZB rollouts on cost.

 

[Annetts key]

On the TPWS installations fed from relay based interlocking that I have worked on (all Western Region free wired interlocking) bar one (which used the separate TPWS signal box unit), in practice, it's been more complex.

The 'fault-monitoring' contact has been wired into the main lamp proving circuit normally by providing an additional relay, a VCR in the relay room for the interlocking because (1), not all signals have a GECPR (this being the Western Region circuit name) and (2), another contact of the HR and if the signal has a subsidiary position light aspect also a contact of the (S)GR is needed in the VCR circuit so that when the TPWS is not energised, that is the signal is showing a proceed aspect, the TPWS 'fault-monitoring' contact being open does not mess up both the signal in rear and the signallers indication for the signal.

I've sketched a typical arrangement. Some signals use repeat relays, so HPR instead of HR for example.


Diagram above shows an example of a typical TPWS Proving Circuit wired into an existing relay interlocking.
A relay, called a VCR is energised when either the signal is showing proceed aspect (either a main aspect, so relay HR is energised or a subsidiary position light aspect, so relay (S)GR is energised) or all TPWS modules (typically a TSS module and a OSS module) have their proving contacts made indicating that they are working "okay".

Also shown is how contacts of the VCR relay are wired into the relay control circuit for the signal in rear (on the approach to the signal that is fitted with TPWS) and how the VCR relay is wired into the circuits that provide the indications for the signal to the signaller.

And just to be clear, to those who are not aware, every single change to a circuit requires a team of installers and a team of testers to work on each change to each circuit of each signal that is affected, including signals in rear that may not even be having a new "safety" system fitted. Hence, this becomes rather expensive when the numbers of signals that are (to be) fitted increases.

== Doublepost prevention - post automatically merged: 3 Jul 2026 ==

The truth is that any train protection system only makes sense if installed everywhere. A collision can happen literally anywhere on a running line as you say. The whole risk-to-cost ratio nonsense that was probably used to conclude that TPWS is only needed at junction signals and select high-risk areas needs to be abandoned completely. If the risk exists and can be mitigated by a train protection system it should be mitigated, no matter whether trains pass through there once or one thousand times a day. I'm almost certain France and Germany didn't base their nationwide KVB/PZB rollouts on cost.

Yes, a collision can happen beyond any signal at red / danger. TPWS was considered lower cost because (1), it was not going to be fitted to every stop signal (signals that can show red or danger) and (2), the modifications to the trains was supposed to be simple and straightforward.

In practice, wiring it into the existing signalling systems was not as easy or as cheap as first expected in many installations. The whole risk-to-cost ratio nonsense may well not have fully taken into account the actual real cost of providing TPWS due to how expensive it is to wire it into existing relay based signalling systems.

And when the trains came in for fitment, it was found that there were problems with the existing AWS on many of them. So that was a problem. Again, the whole risk-to-cost ratio nonsense may well not have fully taken into account the actual real cost of providing TPWS on trains.

And I have no idea how much the on going maintenance and fault rectification with all the delays to trains that these faults cause, costs the industry.

I don't however think that certain lines (for example some branch lines) need urgent fitment of ETCS/ERTMS, if the speed is low, or it's normally only one train on a branch line, or it's only a freight line, the risk is lower or reduced or limited.

But IMHO on all main lines (including "slow" and relief lines) the roll out of ETCS/ERTMS should have been much faster than it has been.

 

[zwk500]

If the risk exists and can be mitigated by a train protection system it should be mitigated

All risk could be prevented by not running trains. Should we do that?

How many accidents have there been which would have been prevented by fitting normal running stop signals with TPWS?

== Doublepost prevention - post automatically merged: 3 Jul 2026 ==

But IMHO on all main lines (including "slow" and relief lines) the roll out of ETCS/ERTMS should have been much faster than it has been.

Completely agree. And NR should have been much more open to L1 only being fitted in areas where renewal/replacement of interlocking was still decades away due to service life left.

 

[rmHawk765]

Completely agree. And NR should have been much more open to L1 only being fitted in areas where renewal/replacement of interlocking was still decades away due to service life left.

Did they ever specifically object to it, or was it mostly down to timescales and money?

How many accidents have there been which would have been prevented by fitting normal running stop signals with TPWS?

One life lost is too many lives lost, it doesn't matter whether there's been accidents like it or not. Seriously, I don't think there's a single other country with a railway system as complex as ours that loves to prioritise train protection infrastructure cost over safety. The reason our railway is so safe is because, thankfully, the safety culture mitigates most of the shortfalls of the infrastructure, and our maintenance is pretty top notch.

But IMHO on all main lines (including "slow" and relief lines) the roll out of ETCS/ERTMS should have been much faster than it has been.

Crucially there are two benefits of an ETCS L2 fitment on a mainline, one of course being safety but the other is capacity, which indeed is the main reason why the East Coast Digital Programme is a thing.

 

[Annetts key]

Crucially there are two benefits of an ETCS L2 fitment on a mainline, one of course being safety but the other is capacity, which indeed is the main reason why the East Coast Digital Programme is a thing.

I think the "selling point" of increased capacity is a bit of a red herring, many lines don't actually run near their existing capacity (assuming a normal undisrupted service).

Since the government decided against the widespread installation of ATP, how many lines have been resignalled? How many new trains have been introduced?

The figures are obviously lower, but since TPWS has been fitted to signals that have a point of conflict ahead of them across the network, how many lines have been resignalled? How many new trains have been introduced?

The can just keeps getting kicked down the track and every time the justification is that we have not killed enough people yet and because of this there is no money for a much better system.

TPWS limits capacity (due to the compromise caused by having to cope with the wide range of train types) and is not very good at enforcing speed limits especially over points and junctions.

And are events (crashes, incidents) that happen / occurred on other railways even taken into account?

The government and industry have squandered time that could have been used for a steady and gradual rollout that had this started many years ago, would have substantially reduced the risk of injuries or deaths on fitted trains travelling on fitted lines.

The technology used in the trackside TPWS equipment was possible and practical in the early 1980s. It's not even classed as a "failsafe" system. We have it because it was seen as being cheap. And because the government at the time that they finally made the decision for TPWS to be fitted nationwide, wanted something quick after two train crashes on the GWML. Yes, I know that TPWS had been in development for years before this.

Meanwhile various other railway networks have fitted train protection systems that are far superior to TPWS.

Meanwhile, in comparison, this country has used large amounts of money to use technology to make so called 'smart' motorways which has resulted in them being more dangerous should your vehicle breakdown. Please note I don't want to take this off topic to discuss roads.

Crucially there are two benefits of an ETCS L2 fitment on a mainline, one of course being safety but the other is capacity, which indeed is the main reason why the East Coast Digital Programme is a thing.

Also I dislike this use of the word digital as if it's something wonderful. Morse code transmitted by ship using light signals is digital... And how long has that been possible?

Most of the traditional signalling systems are digital - the signal is either showing one of a number of defined colours otherwise it's faulty. It's certainly not an analogue system. The East Coast signalling system was already a digital system.

What they really mean is that it's a computer based signalling, control supervision and safety system.

 

[Bald Rick]

One life lost is too many lives lost, it doesn't matter whether there's been accidents like it or not. Seriously, I don't think there's a single other country with a railway system as complex as ours that loves to prioritise train protection infrastructure cost over safety. The reason our railway is so safe is because, thankfully, the safety culture mitigates most of the shortfalls of the infrastructure, and our maintenance is pretty top notch.

You need to read up on train protection systems elsewhere in the world. And see how many head on / rear end collissions there are in, say, Germany, Austria, Belgium, etc.

 

[HSTEd]

I also haven't been able to find any illustration of the actual equipment! I expect they'll be fairly compact, with a serial data connection from LEU to a radio transceiver and antenna.

I have been unable to find a picture despite more searching, but as far as I can tell it is just a GSM-(R) telephone. The train essentially dials a phone number provided by a balise for updates.

Given that GSM-R handsets exist I think it would be possible to build a combined LEU/RIU that can control one or two balises and also act as a radio infill unit. With monitored current transformer connections it should be possible to "drop" them in to return signal aspects.

Each section would have a transparent balise that gives the signal aspect and also the RIU number for the section beyond the signal.

Most banner repeaters and the like can simply be removed, although I imagine some quipment has to remain to keep the lamp proving circuitry happy.

EDIT:

I have located one! Apparently made by a company called 'Funkwerk' - 4 radios fit into a 3U 19" rack unit.
So they are very compact, these particular ones appear to use RS-422 and RS-232 to communicate.

The rack with four radios is 132mm tall, 190mm deep and 19"/483mm across. They can even come with integrated data loggers.

RIU-ETCS - Funkwerk's mobile solution for ETCS applications consists in the standard version of fully independent radio units EDOR-5E with integrated state-of-the-art GSM-R mobile radio module MT5-E (8 Watt) and the wide-range power supply (24 to 110 V on-board voltage). The EDOR-5E fulfils the latest specification for improved receiver parameters ETSI TS 102 933 V2.1.1.

 

[Annetts key]

Most banner repeaters and the like can simply be removed, although I imagine some quipment has to remain to keep the lamp proving circuitry happy.

Either the relevant contacts can be removed from the circuit and the circuit rewired slightly. Or if a repeat relay is used for the lamp proving, this can be fed via a constant supply, hence keeping the relay energised all the time. So this is not difficult.

 

[HSTEd]

Apparently the National Rail system has ~40,000 signals, whereas each "National Identifier" (NID_C, a ten bit binary number) can only contain one set of 14-bit designations for Radio Infill Units (RIU), which means only 16,384 Radio Infill Units can be configured on any given National Network identifier code. Although I'm not sure what fraction of the signals are actual stop signals (as opposed to banners and distants) that would require RIUs for protection/cab signalling.

We have about 40,000 signals across the whole network, mostly colour light signals and with many benefitting from LED technology. There are also still some traditional mechanical semaphore signals.

Luckily, for some reason the UK has been assigned 50 NID_C codes...... (pg 9/96 of the PDF), which is nearly 5% of the total!

So it should be possible to fit RIUs to every single home/stop signal on the railway with plenty to spare if that is necessary.

 

[Annetts key]

Apparently the National Rail system has ~40,000 signals, whereas each "National Identifier" (NID_C, a ten bit binary number) can only contain one set of 14-bit designations for Radio Infill Units (RIU), which means only 16,384 Radio Infill Units can be configured on any given National Network identifier code. Although I'm not sure what fraction of the signals are actual stop signals (as opposed to banners and distants) that would require RIUs for protection/cab signalling.

The total number of signals may include:
Semaphore stop signals (home, starter etc.),
A small number of semaphore distant signals (including fixed at caution types),
Stop boards,
Fixed distant boards,
Mechanical ground shunting signals,
Fixed at red colour light signals (a single red aspect head),
Two aspect red/green colour light stop signals,
Two aspect yellow/green colour light distant/repeater signals,
Three aspect yellow/double yellow/green colour light distant/repeater signals,
Three aspect red/yellow/green colour light signals,
Four aspect red/yellow/double yellow/green colour light signals,
Banner signals (whatever type/technology),
Ground position light signals (GPL), including Limit Of Shunt (LOS) types (fixed at red),
Maybe some traditional LOS types,
Elevated (i.e. on a post) position light signals.

I don't know if Network Rail are including the following:
Level crossing drivers white light (now drivers red/white light),
Level crossing Barriers Up (BU) illuminated indicator signal,
Toton shunting signals,
Elevated single white aspect shunting signals (repeat a GPL).

Not all of these are stop signals.

 

[MarkyT]

Luckily, for some reason the UK has been assigned 50 NID_C codes...... (pg 9/96 of the PDF), which is nearly 5% of the total!

So it should be possible to fit RIUs to every single home/stop signal on the railway with plenty to spare if that is necessary.

Countries seem to have been assigned blocks of NID_C codes proportional to their network size. Germany has over 60 for example, whereas Belgium has only 5.

To demonstrate the UK are not far behind on ETCS rollout, only a handful of pilot ETCS schemes are in use on DB currently, including some L1 LS examples, and the Berlin S-Bahn has a custom packet 44 system that replaced mechanical train stops with balises. The difference is the level of protection provided by the legacy PZB system, which is almost universal across all main signals on the network today.

Here's an interesting article from 2014 detailing the Swiss approach to 'industrialised' Level 1 LS fitting across the entire non-Level 2 network, via a short-lived conversion of their legacy Signum/ZUB transponders to balises.

https://www.railweb.ch/2014-04_IRSE_News_ETCS_L1LS.pdf

Siemens'Rollout of ETCS L1 Limited Supervision on 5BB
Siemens opened the doors of their Wallisellen site on 22 November to show the IRSE Swiss Section the rollout of ETCS Level 1 Limited Supervision on Swiss Federal Railways (SBB). Thomas Habermacher and Thomas O§erholzer explained how innovations in products but also in the industrial processes underlying the five-year rollout, are drastically cutting conversion costs.

I think the "selling point" of increased capacity is a bit of a red herring, many lines don't actually run near their existing capacity (assuming a normal undisrupted service).

Agreed. On very busy lines with blocks shorter than typical colour lights, the technology can allow quicker reoccupation at platforms, with block markers close to the running-in ends, like closing up signals. These were provided selectively in colour light signalling layouts, of course, but where through speed is also high, aspect sequences often become rather complex and confusing with repeated yellows and timed releases where there's not braking distance between particular signals. With no need for long-range optical visibility, the cost of major signalling structures across multiple tracks could also be reduced on new schemes.

Shorter blocks also limit how much capacity reduction there is when temporary speed restrictions apply. Spacing of conventional signals is 'tuned' to an optimum speed. If you run through a fixed block section at half the usual speed, it takes twice as long to clear it, affecting following headway.

Level 2 doesn't have to provide more capacity; Markers could be placed only at the same positions as existing signals. Marker board spacing isn't tied to braking distance and can be variable however, so additional blocks can be squeezed in more easily where advantageous, particularly on approach to complex stations and junctions.

Since the government decided against the widespread installation of ATP, how many lines have been resignalled? How many new trains have been introduced.

The figures are obviously lower, but since TPWS has been fitted to signals that have a point of conflict ahead of them across the network, how many lines have been resignalled? How many new trains have been introduced?

It's very difficult to justify doing any typically sized signalling renewals project with ETCS L2 today. Most fleets are still unfitted, and each scheme would become a tiny island still surrounded by a great sea of colour lights.

The can just keeps getting kicked down the track and every time the justification is that we have not killed enough people yet and because of this there is no money for a much better system.

What's inevitable is that the remaining risk becomes concentrated on the signals that are still unfitted.

In Belgium the 2016 Hermalle-sous-Huy collision occurred when a passenger train ran through a plain line danger signal and hit a slow-moving freight in front, killing 3. The signal concerned was one of the last handful still awaiting the TBL1+ system. Belgium modernised their TBL1 system, with GW-ATP style beacons and French style Crocodiles, to use Eurobalises. They also decided to equip all signals with protection. The previous version was only provided at the highest risk locations, rather like TPWS. ISTR the equipment at the accident site was due to be commissioned imminently.

TPWS limits capacity (due to the compromise caused by having to cope with the wide range of train types) and is not very good at enforcing speed limits especially over points and junctions.

It also can't prevent reacceleration after passing successfully through an overspeed trap under the set speed. AWS can't enforce a slowdown after a caution once it has been acknowledged. PZB does both.

And are events (crashes, incidents) that happen / occurred on other railways even taken into account?

The government and industry have squandered time that could have been used for a steady and gradual rollout that had this started many years ago, would have substantially reduced the risk of injuries or deaths on fitted trains travelling on fitted lines.

The rollout to universal L2 is not economic. We need cheaper systems for regional routes, probably 75% of UK route mileage.

The technology used in the trackside TPWS equipment was possible and practical in the early 1980s. It's not even classed as a "failsafe" system. We have it because it was seen as being cheap. And because the government at the time that they finally made the decision for TPWS to be fitted nationwide, wanted something quick after two train crashes on the GWML. Yes, I know that TPWS had been in development for years before this.

It was also easy and quick to apply, which was very important in the circumstances. There really was little else available off the shelf, and no large railway organisation at that time would deliberately shackle themselves to a new proprietary full-fat ATP system when ETCS was just over the horizon.

Meanwhile various other railway networks have fitted train protection systems that are far superior to TPWS.

Many of them predated even BR-AWS. I consider German PZB or 'indusi' to be the gold standard, which originated in the 1930s. In its latest form 'PZB90' has become near-universal in Germany, though was confined to the busiest main lines and rail hubs for many years until a government intervention forced its wider rollout following several serious collision incidents.

Swiss Signum is a related system with an inductive trainstop, though didn't have the 1000Hz equivalent overspeed magnets on final approach. The ZUB system was invented for infill using a leaky feeder track loop over ~300m on approach at selected signals.

The UK was very late on the scene for the kind of trainstop and overspeed functionality pioneered in PZB.

 

[andy33gmail]

Shorter blocks also limit how much capacity reduction there is when temporary speed restrictions apply. Spacing of conventional signals is 'tuned' to an optimum speed. If you run through a fixed block section at half the usual speed, it takes twice as long to clear it, affecting following headway.

So in a fixed block system, to an extent, shorter blocks increase throughput;

In view of your ‘tuning’ observation; does it follow that in some circumstances, lower speed limits improve headway - either by better matching block sizes to stopping distances or allowing more blocks?

 

[zwk500]

So in a fixed block system, to an extent, shorter blocks increase throughput;

Yes. With conventional signalling, minimum block length is constrained by train braking performance. With ETCS L2, that constraint is removed (one of the main capacity benefits). However, there is a cost associated with more blocks.

In view of your ‘tuning’ observation; does it follow that in some circumstances, lower speed limits improve headway - either by better matching block sizes to stopping distances or allowing more blocks?

Yes. Increasing linespeed can require signals to be spaced further apart, so there's a balance for headway.

 

[andy33gmail]

Yes. With conventional signalling, minimum block length is constrained by train braking performance. With ETCS L2, that constraint is removed (one of the main capacity benefits). However, there is a cost associated with more blocks.

Yes. Increasing linespeed can require signals to be spaced further apart, so there's a balance for headway.

Is that because conventional signals only have 2-4 aspects, so maximum stopping distance from line speed needs to be attainable within 3 blocks?

 

[MarkyT]

So in a fixed block system, to an extent, shorter blocks increase throughput;

This simulation in Transport Fever 2 illustrates the effect of a severe speed restriction through a fixed block spaced for normal line speed. Splitting the block helps prevent the further delay rippling back into oncoming traffic.

Yes. Increasing linespeed can require signals to be spaced further apart, so there's a balance for headway.

Also, at lower speeds, while braking Distance reduces, the fixed constants of Sighting distance, Overlap and train Length start to dominate the headway equation.
The headway distance is calculated thus:

4 aspect H4=S+1.5D+O+L
3 aspect H3=S+2D+O+L

The technical headway time separation is simply the time taken to cover that distance at the speed being used to calculate D.

Table A (all trains) braking distance on the level varies between 2041m at 100mph and 325m at 25m. Placing signals only a little more than a train length apart also risks a forward train stopping still foul of the overlap of the previous signal, preventing a close following approach. Note overlaps can be reduced from their default 183m at lower approach speeds, and sighting can be deliberately reduced by specifying short-range heads where there's a risk of read-through for closely spaced signals.

Is that because conventional signals only have 2-4 aspects, so maximum stopping distance from line speed needs to be attainable within 3 blocks?

Yes, 4-aspect is the shortest block solution we have in conventional UK TCB colour light signalling, though an additional signal or two can be introduced approaching stations, for example, to allow the next train to 'close up' nearer the platform awaiting the previous departure.

Approaching termini, the speed limit can often be managed down to achieve this with a blanket restriction across a large throat, for example, without any weird aspect sequences, but applications can be complex if other trains also pass through a station at speed, as special aspect sequences may need to be devised that can carry their own risks of anticipation error.

There must always be braking distance from the first caution to the red.

 

[bahnause]

The rollout to universal L2 is not economic. We need cheaper systems for regional routes, probably 75% of UK route mileage.

Yet another case where the seller has ripped off the buyer. ETCS L2 was always touted as the most cost-effective option, as doing away with external signalling and using standardised hardware on the trains was supposed to make everything much better, cheaper and more compatible. In reality, none of that has been achieved.

One life lost is too many lives lost, it doesn't matter whether there's been accidents like it or not. Seriously, I don't think there's a single other country with a railway system as complex as ours that loves to prioritise train protection infrastructure cost over safety. The reason our railway is so safe is because, thankfully, the safety culture mitigates most of the shortfalls of the infrastructure, and our maintenance is pretty top notch.

I don’t think there is a railway in the world that does not weigh up safety, risks and availability against one another. Even with the relatively new ETCS L1LS system in Switzerland, the monitoring functions are still carried out on a risk-based basis and are therefore not always at the maximum available safety level. At many signals, just as 50 years ago, monitoring is limited to ‘Stop’ and ‘Warning’ signals, even though, technically speaking, braking curve monitoring would be possible. And anyone who has ever carried out such planning will understand why this is done. The planning effort involved is enormous; for more complex stations, we estimate planning periods of over one year for a full-scale installation.

 

[rmHawk765]

At many signals, just as 50 years ago, monitoring is limited to ‘Stop’ and ‘Warning’ signals,

And TPWS fails to achieve warning functionality, not to mention stop functionality not being present at all signals which is precisely what this thread is about...

 

[bahnause]

And TPWS fails to achieve warning functionality, not to mention stop functionality not being present at all signals which is precisely what this thread is about...

Which is the important bit. On our infrastructure with only one function, it is "Warning" and not "Stop".

 

[HSTEd]

Beyond saturation train protection, I think cab signalling is potentially very useful if it can be deployed at non-stunning cost.

Once safety grade signalling information can be provided to the train computer using ETCS, deploying ATO for improved operational efficiency (or even moving towards GoA3) is much simpler.

I could envisage a ATO-capable cab signalling system that uses ETCS Level 1-type movement authorities (conveyed by transparent balises and radio infill units) being combined with a GSM-R data channel for transmitting the other non-safety grade information required for ATO to the train. That information would include "what stations am I stopping at" and train regulation information.

I don't think it really matters how the ETCS datagrams are derived, as long as they are all valid datagrams. My understanding is that the signalling industry is having issues keeping up with the required rate of resignalling as it is, so if we could repurpose existing interlockings that could be useful.

As I said earlier, conceptually we could go so far as having a Grade of Automation (GoA) 3 (ie. train crew on board but not in the cab) train being controlled by a signaller in a lever frame box.

ETCS Level 2 is cool, but the Radio Block Computer concept does seem to be a major break with conventional practice.

 

[Belperpete]

Either the relevant contacts can be removed from the circuit and the circuit rewired slightly. Or if a repeat relay is used for the lamp proving, this can be fed via a constant supply, hence keeping the relay energised all the time. So this is not difficult.

I am not sure that a traditional electro-mechanical banner signal can be easily removed. Isn't it proved on in the signal on-proving, and proved off in the higher aspects of signals in rear? Pretending that the banner is both on and off could be interesting. But removing a modern LED banner should as you say be easy, if you false-feed the lamp-proving in some way.

I can recall the relay room at Waterloo (or might have been Victoria, back in L frame days) having a number of signal heads inside it merrily working away, changing aspects every now and then, in place of mid-platform signals that had never been commissioned, but were required for the interlocking to work.

== Doublepost prevention - post automatically merged: 5 Jul 2026 ==

On the TPWS installations fed from relay based interlocking that I have worked on (all Western Region free wired interlocking) bar one (which used the separate TPWS signal box unit), in practice, it's been more complex.

The 'fault-monitoring' contact has been wired into the main lamp proving circuit normally by providing an additional relay, a VCR in the relay room for the interlocking because (1), not all signals have a GECPR (this being the Western Region circuit name) and (2), another contact of the HR and if the signal has a subsidiary position light aspect also a contact of the (S)GR is needed in the VCR circuit so that when the TPWS is not energised, that is the signal is showing a proceed aspect, the TPWS 'fault-monitoring' contact being open does not mess up both the signal in rear and the signallers indication for the signal.

Only buffer stops and separate speed restriction OSS were wired into the first filament failure circuit (on Western this being the G(M)ESR circuit).

Additionally, in some installations extra special controls had to be provided, because of either where the track circuits (IRJs or the turned zone for ASTER U/SF15/TI21) were, where it was possible to actually fit the TPWS TSS loops ('racks') because of either points or an existing ATP signal beacon or other reasons, otherwise, the signal could go back to red as the front of the train passed it, which caused the TPWS to energise, which then tripped the brakes on the train because the receiver under the train was still in range of the track mounted loops ('racks') causing the driver to be a bit confused because they had only seen a proceed aspect...

Indeed, the very first two TPWS signal installations in the area where I was, were taken out of use after being in service for only one day because of problems. It was not until a solution was found and implemented that these two installations were brought back into use.


This may have been the case for some installations, but it was not the same everywhere. Lots of relay rooms (mostly those that contained the local interlocking) had to have extra relays installed and extra 12 core cables (to the new TPWS cases) and some TPWS modules were fitted either in relay rooms or very occasionally in existing location cupboards.

That's the problem with trying to come up with a simple low cost solution, there isn't a standard colour light signal installation. So you can't have a standard universal new system which wires into the existing signalling system. The nearest that I have seen is how the GWML ATP uses current transformers made to fit on existing 0BA or 2BA link strips. Even that needed one extra circuit for junction signals if one route did not have a junction or route indicator.

I was talking about the original TPWS concept.

Simply monitoring the voltage across the red aspect could have been applied to just about all UK technologies - just connect a pair of wires across the red aspect feed. Likewise cutting the proving into the filament-proving circuit would just have needed a link removing to break the circuit, and a pair of wires connected across it. Likewise could have been implemented with just about any technology. And like some of the condition monitoring equipment that has been overlaid to many level crossings and other lineside equipment, would have been installed with a minimal design and testing requirement.

Unfortunately by the time it got through safety approval, it had morphed into something significantly different. I don't think they could have come up with something more difficult to interface with SSI if they had deliberately set out to do so! The TPWS equipment itself was (I think) the same, but the means of interfacing it to the signals had been totally changed. Gone was the simple two wires across the red aspect, it now required interface relays. Gone was the simple cut into the filament-proving circuit, it now had to be cut into the lamp-proving, which added timing issues that bedevil TPWS to this day. And this of course meant alterations to existing safety circuits requiring the full gamut of design, check and test requirements. At every signal to be fitted.

I understand that it was considered that the SIL level wasn't high enough - even though the whole concept was that it wasn't a fail safe system. So, for example, it was decided that it was unacceptable for a train to run at speed up to a red signal with a failed TPWS - the train has to be cautioned up to the "failed signal", in the same way as if the red lamp has gone out. Hence the TPWS proving having to be cut into the signal lamp-proving rather than the filament proving.

Anyone who thinks they have a simple concept for a new system, or enhancement to the existing, should take a long hard look at what happened with TPWS.

== Doublepost prevention - post automatically merged: 5 Jul 2026 ==

Yet another case where the seller has ripped off the buyer. ETCS L2 was always touted as the most cost-effective option, as doing away with external signalling and using standardised hardware on the trains was supposed to make everything much better, cheaper and more compatible. In reality, none of that has been achieved.

The problem in the UK is that, with TPWS having been rolled out, the financial case for a proper ATP system is now much reduced. The number of lives that will be saved by fitting ATP is much less than would have been the case without TPWS. But ATP will still cost the same to install, so the cost in millions of pounds to save each life is significantly higher. However, whenever there is an accident that ATP would have prevented, there is still a clamour (as on this thread) to do something. Hence the push to promote ATP's other benefits. And undoubtedly those benefits are being over sold.

I don’t think there is a railway in the world that does not weigh up safety, risks and availability against one another. Even with the relatively new ETCS L1LS system in Switzerland, the monitoring functions are still carried out on a risk-based basis and are therefore not always at the maximum available safety level. At many signals, just as 50 years ago, monitoring is limited to ‘Stop’ and ‘Warning’ signals, even though, technically speaking, braking curve monitoring would be possible. And anyone who has ever carried out such planning will understand why this is done. The planning effort involved is enormous; for more complex stations, we estimate planning periods of over one year for a full-scale installation.

Even TPWS involves significant planning costs. One glance at a recent signalling plan will show rafts of TPWS calculations. Every signal, speed restriction and buffer stop has to have a risk assesment as to whether to fit TPWS or not. Assesments made as to whether OSS, OSS+, OSS++, OSS- loops are required. Then the theoretical position of each loop, it's maximum and minimum release speeds, and loop spacings all have to be calculated - and then adjusted for practical site conditions. And not to mention the design effort to interface it with the signalling equipment. And all that long before anyone goes out with a loop and toolkit in hand (having first done a risk assesment of the site and obtained a possession of the line, of course).

 

[MarkyT]

I am not sure that a traditional electro-mechanical banner signal can be easily removed. Isn't it proved on in the signal on-proving, and proved off in the higher aspects of signals in rear? Pretending that the banner is both on and off could be interesting. But removing a modern LED banner should as you say be easy, if you false-feed the lamp-proving in some way.

I expect the electro-mechanical variety are few and far between today.

Simply monitoring the voltage across the red aspect could have been applied to just about all UK technologies - just connect a pair of wires across the red aspect feed. Likewise cutting the proving into the filament-proving circuit would just have needed a link removing to break the circuit, and a pair of wires connected across it. Likewise could have been implemented with just about any technology. And like some of the condition monitoring equipment that has been overlaid to many level crossings and other lineside equipment, would have been installed with a minimal design and testing requirement.

Unfortunately by the time it got through safety approval, it had morphed into something significantly different. I don't think they could have come up with something more difficult to interface with SSI if they had deliberately set out to do so! The TPWS equipment itself was (I think) the same, but the means of interfacing it to the signals had been totally changed. Gone was the simple two wires across the red aspect, it now required interface relays. Gone was the simple cut into the filament-proving circuit, it now had to be cut into the lamp-proving, which added timing issues that bedevil TPWS to this day. And this of course meant alterations to existing safety circuits requiring the full gamut of design, check and test requirements. At every signal to be fitted.

I thought in original retrofits to SSI, the signal voltage triggering was the same, but I can see the lamp proving being more complex as most early SSI installations used internal return current proving in the module, so you'd need to allocate a new input for TPWS healthy and change the data.

I understand that it was considered that the SIL level wasn't high enough - even though the whole concept was that it wasn't a fail safe system. So, for example, it was decided that it was unacceptable for a train to run at speed up to a red signal with a failed TPWS - the train has to be cautioned up to the "failed signal", in the same way as if the red lamp has gone out. Hence the TPWS proving having to be cut into the signal lamp-proving rather than the filament proving.

The fundamental problem is that in a power failure the equipment is completely dead and is not seen at all by the trainborne equipment. By contrast, AWS can rely on its permanent magnet giving a warning signal in a power outage. It absolutely had to be closely monitored directly in approaching signal circuits rather than just generating a signaller or technician zone alarm that might be one of a dozen lamps or grids.

Anyone who thinks they have a simple concept for a new system, or enhancement to the existing, should take a long hard look at what happened with TPWS.

It was a hard slog for all those involved, but they completed the project very quickly, though it seemed at one time that every designer was doing TPWS work. There was a similar story more recently in the USA, completing PTC fitment over the network. I don't doubt in the 90s, when the decision to fit all signals in Germany with PZB was taken, the industry there faced similar challenges.

The problem in the UK is that, with TPWS having been rolled out, the financial case for a proper ATP system is now much reduced. The number of lives that will be saved by fitting ATP is much less than would have been the case without TPWS. But ATP will still cost the same to install, so the cost in millions of pounds to save each life is significantly higher. However, whenever there is an accident that ATP would have prevented, there is still a clamour (as on this thread) to do something. Hence the push to promote ATP's other benefits. And undoubtedly those benefits are being over sold.

It's mainly the assumption that full-fat mainline Level 2 is the universal solution to all things from busy urban to rural branch and all the emerging costs and difficulties of those systems. Most European countries have at least research projects, trials or pilots involving Level 1 LS as a replacement for legacy Class B national system on a large proportion of their regional networks, usually to supplement their busy trunk Level 2 lines and extend interoperability. Some have gone further; Belgium and Switzerland were notable early adopters; France's KVB, while not directly compatible, is the OG balise-based LS system, and eminently convertible from the same signalling inputs though that'd be a lot of new balises; Germany is looking at a ETCS L1 LS based successor to PZB, and Poland, as mentioned upthread, is preparing for widespread implementation, perhaps sooner than DB as the Polish legacy system is less secure. Great Britain remains a notable exception, along with Denmark, perhaps among others who remain true to L2 alone. The difference in delaying implementation here, compared to, say, Germany, is the level of protection provided by the legacy systems.

 

[bahnause]

It's mainly the assumption that full-fat mainline Level 2 is the universal solution to all things from busy urban to rural branch and all the emerging costs and difficulties of those systems. Most European countries have at least research projects, trials or pilots involving Level 1 LS as a replacement for legacy Class B national system on a large proportion of their regional networks, usually to supplement their busy trunk Level 2 lines and extend interoperability. Some have gone further; Belgium and Switzerland were notable early adopters; France's KVB, while not directly compatible, is the OG balise-based LS system, and eminently convertible from the same signalling inputs though that'd be a lot of new balises; Germany is looking at a ETCS L1 LS based successor to PZB, and Poland, as mentioned upthread, is preparing for widespread implementation, perhaps sooner than DB as the Polish legacy system is less secure. Great Britain remains a notable exception, along with Denmark, perhaps among others who remain true to L2 alone. The difference in delaying implementation here, compared to, say, Germany, is the level of protection provided by the legacy systems.

This assumption that full-scale Level 2 is the universal solution stems from the fact that ETCS Level 2 was touted precisely for that purpose. However, it turned out to be a complete failure, both economically and operationally. ETCS is not cost-effective, regardless of the level implemented. Anyone who wants a cost-effective solution – and is permitted to do so under the regulations – will opt out of ETCS. The entire non-interoperable network in Switzerland is not being fitted with ETCS, but with infrastructure-specific systems like ZBMS. Through ZBMS, Switzerland successfully brought modern, high-tier digital safety monitoring to its narrow-gauge lines without the unnecessary financial overhead and technical constraints of standard ETCS. Equipping the network with ETCS is not a matter of choice, but is being carried out in accordance with government requirements. Those who are not obliged to do so do not use it in the intended form, but at best use parts of the system or the hardware.

DB uses L1LS only for cross-border services; there are no further projects and there will be no further routes fitted with L1LS. Switzerland only rolled out L1LS because the costs for L2 were too high and they wanted to avoid having to retrofit the doors of numerous vehicles. After years of deliberation by the regulatory authorities and infrastructure operators on how to get out of this mess, it has become clear that there is no way out. Consequently, network-wide L2 is the new and current strategy. However, the system needs to be optimised, as ETCS Level 2 is a major drain on capacity on the Swiss network.

The Federal Office of Transport (FOT) and SBB have now, albeit grudgingly, recognised this “capacity drain”. The initial aim of blindly rolling out Level 2 across the board has been halted. The current counterstrategy is as follows:

• Optimisation of braking curves: Intensive work is underway on software updates (e.g. optimisations in ETCS Baseline 3 and 4) to teach the system “Swiss driving behaviour”, so that trains can once again decelerate more dynamically.
• Demand-driven roll-out: At highly complex, congested stations, Level 2 is being held back for the time being until the software is mature enough not to compromise capacity.

 

[Belperpete]

It's very difficult to justify doing any typically sized signalling renewals project with ETCS L2 today. Most fleets are still unfitted, and each scheme would become a tiny island still surrounded by a great sea of colour lights.

The rollout to universal L2 is not economic. We need cheaper systems for regional routes, probably 75% of UK route mileage.

The rollout of any other ATP system, be it L1 or something else, is going to need the fleets fitted. Surely it is more economical to fit them with one system, that can be used anywhere, rather than fit them with two different systems, as most stock will inevitably run on both regional routes and main lines at some point, requiring dual-fitted stock and/or islands of dual-fitted signalling. Retrofitting stock with a second form of ATP is surely going to make the problem worse, not better.

The advantage I see of L2 is that the interface is at interlocking level, between the interlocking and ETCS RBC. Yes, it needs a swathe of new interlocking data for the interface, but it doesn't need an army of technicians and testers going out trackside to interface balises to trackside equipment cases (the balises are purely passive position markers).

I agree with what you say about the problem of spot resignalling with L2. Currently such installations need to be dual-signalled, as so little stock is ETCS fitted. Dual signalling inevitably costs more (significantly more) than just providing colour-lights. The solution surely is to bit the bullet and get the stock, particularly freight stock, fitted. Once that is done, then resignalling schemes will no longer need the colour lights, aws and tpws. The savings from not having to provide all that trackside kit should surely outweigh the extra interlocking data, RBC, etc.

This was all recognised in the ETCS position papers, probably a decade or more ago. You have to go all out for ETCS, with an aligned rollout of both signalling and train fitment. You also have to accept that much of the early fitment is more expensive, with more expensive dual signalling in some cases, and stock being fitted in advance of it really being needed. Attempting to roll it out in an uncoordinated fashion will just mean it continues to be more expensive.

Unfortunately we seem to have forgotten that. For example, surely the new class 197s should ALL have been ETCS fitted, not just some of them. Unfortunately in the UK, short term savings always win, regardless of the long term implications. We buy cheap, and pay the consequences over the years. I fear that the same will happen with ETCS.

I do not see the implementation of a second system, particularly one that needs loads of trackside interfaces, as the solution.

 

[MarkyT]

This assumption that full-scale Level 2 is the universal solution stems from the fact that ETCS Level 2 was touted precisely for that purpose. However, it turned out to be a complete failure, both economically and operationally. ETCS is not cost-effective, regardless of the level implemented. Anyone who wants a cost-effective solution – and is permitted to do so under the regulations – will opt out of ETCS. The entire non-interoperable network in Switzerland is not being fitted with ETCS, but with infrastructure-specific systems like ZBMS. Through ZBMS, Switzerland successfully brought modern, high-tier digital safety monitoring to its narrow-gauge lines without the unnecessary financial overhead and technical constraints of standard ETCS. Equipping the network with ETCS is not a matter of choice, but is being carried out in accordance with government requirements. Those who are not obliged to do so do not use it in the intended form, but at best use parts of the system or the hardware.

The mainline operators need to look closely at ZBMS, which, according to this article, is at heart a level 1 LS system using many ETCS components and protocols. The main difference operationally is a lack of cab signals, suggesting it monitors silently in the background until an intervention, like TPWS.
Edit: thinking about that it's genius, especially on light rural unfenced railways through challenging terrain where you want the driver in full situational awareness looking out of the window for any upcoming hazards, not interacting with screens.

https://www.railweb.ch/2015-10_IRSE_News_Non-interoperable_Railways.pdf

DB uses L1LS only for cross-border services; there are no further projects and there will be no further routes fitted with L1LS. Switzerland only rolled out L1LS because the costs for L2 were too high and they wanted to avoid having to retrofit the doors of numerous vehicles. After years of deliberation by the regulatory authorities and infrastructure operators on how to get out of this mess, it has become clear that there is no way out. Consequently, network-wide L2 is the new and current strategy. However, the system needs to be optimised, as ETCS Level 2 is a major drain on capacity on the Swiss network.

I noted there was no declared end date for phasing out PZB, unlike some other networks and their legacy systems, while PZB is still strongly supported by manufacturers, and many of the cross-border ETCS projects will remain dual-fitted with both systems in operation.

The Federal Office of Transport (FOT) and SBB have now, albeit grudgingly, recognised this “capacity drain”. The initial aim of blindly rolling out Level 2 across the board has been halted. The current counterstrategy is as follows:

• Optimisation of braking curves: Intensive work is underway on software updates (e.g. optimisations in ETCS Baseline 3 and 4) to teach the system “Swiss driving behaviour”, so that trains can once again decelerate more dynamically.
• Demand-driven roll-out: At highly complex, congested stations, Level 2 is being held back for the time being until the software is mature enough not to compromise capacity.

This is good news. Hopefully lessons learned will also be noted by other national implementations.

 

[HSTEd]

The mainline operators need to look closely at ZBMS, which, according to this article, is at heart a level 1 LS system using many ETCS components and protocols. The main difference operationally is a lack of cab signals, suggesting it monitors silently in the background until an intervention, like TPWS.

ETCS Level 1 does appear to have an astonishing capability for adaption to local conditions that ETCS Level 2 simply does not have.

Although is there much point installing an ATP system that, almost by definition, has to know what the speed limit and distance-to-stop is, and then not using it as a cab signalling interface?

Even if we only rip out repeaters and such the savings in equipment could be substantial.

With RIUs, which are probably rather cheap in bulk given their GSM heritage, we could remove everything except for (probably) shunting signals.

 

[Belperpete]

This assumption that full-scale Level 2 is the universal solution stems from the fact that ETCS Level 2 was touted precisely for that purpose. However, it turned out to be a complete failure, both economically and operationally. ETCS is not cost-effective, regardless of the level implemented. Anyone who wants a cost-effective solution – and is permitted to do so under the regulations – will opt out of ETCS.

A big problem in the UK is that doing work trackside is extremely constrained, you effectively need a full line possession to do most trackside work. Getting such possessions is difficult, and hence anything needing trackside work (such as to connect balises to trackside equipment cases) becomes very time consuming and expensive. Thankfully TPWS was rolled out before things became so restrictive, god knows how long the TPWS rollout would have taken and how much it would have cost under today's constraints.

I understand that ETCS L2 is being rolled out between Heathrow and Paddington relatively easily (and I accept that the relatively there is doing some heavy lifting!), because it requires relatively little trackside work. The interface between the ETCS and the conventional signalling is being done at interlocking level. That is the big advantage of L2. You don't need to do work trackside to retrofit it, other than just fit simple dumb positioning balises.

The big disadvantage is that you need an ETCS compatible interlocking to do it. Fortunately in the UK, our national standard interlocking SSI is compatible with ETCS (okay, you have to replace the actual interlocking cubicle with a more modern cubicle, but it runs the same data and talks to the same trackside equipment). And fortunately the line between Heathrow and Paddington was signalled with SSI.

I can well understand that countries and railways where the existing interlockings are not ETCS compatible have a much bigger problem. They will probably have to do a full resignalling to fit L2. And if it is also much easier for them to fit active trackside equipment, such as for L1, then the balance of cost effectiveness probably swings differently.

== Doublepost prevention - post automatically merged: 5 Jul 2026 ==

ETCS Level 1 does appear to have an astonishing capability for adaption to local conditions that ETCS Level 2 simply does not have.

Please explain.

If you mean that it can be much more easily interfaced to existing non ETCS compatible signalling systems, then I would agree. But that comes at a cost, of every signal needing its own local interface. In the UK, such retrofitting would these days come at a very high cost, high disruption, and probably long timescales, due to the constraints on trackside working.

 

[MarkyT]

A big problem in the UK is that doing work trackside is extremely constrained, you effectively need a full line possession to do most trackside work. Getting such possessions is difficult, and hence anything needing trackside work (such as to connect balises to trackside equipment cases) becomes very time consuming and expensive. Thankfully TPWS was rolled out before things became so restrictive, god knows how long the TPWS rollout would have taken and how much it would have cost under today's constraints.

I understand that ETCS L2 is being rolled out between Heathrow and Paddington relatively easily (and I accept that the relatively there is doing some heavy lifting!), because it requires relatively little trackside work. The interface between the ETCS and the conventional signalling is being done at interlocking level. That is the big advantage of L2. You don't need to do work trackside to retrofit it, other than just fit simple dumb positioning balises.

The big disadvantage is that you need an ETCS compatible interlocking to do it. Fortunately in the UK, our national standard interlocking SSI is compatible with ETCS (okay, you have to replace the actual interlocking cubicle with a more modern cubicle, but it runs the same data and talks to the same trackside equipment). And fortunately the line between Heathrow and Paddington was signalled with SSI.

I can well understand that countries and railways where the existing interlockings are not ETCS compatible have a much bigger problem. They will probably have to do a full resignalling to fit L2. And if it is also much easier for them to fit active trackside equipment, such as for L1, then the balance of cost effectiveness probably swings differently.

It's not just other countries. The UK still has a lot of non-processor based signalling. When finances are tough, more peripheral places on the network often have their renewals deferred and a programme of life extension occurs instead. Line of route based programmes thus fall apart as the renewals funding component of constituent projects evaporates.

The tacit policy on fleet has been to let most older fleets die out while their replacements are at least designed to be ETCS ready if not all fitted from new. If e-TPWS/AWS with some extra safety features and wider fitment was available and used, that would perhaps persuade train owners and operators to equip more of their fleets with at least this basic ETCS functionality, which might be upgraded later to full L2 capability.