Is a new system needed to automatically stop trains which pass through a red signal?

[Bletchleyite]

Yes, PZB (as Indusi was renamed) is a remarkable achievement in and of itself, especially considering how long ago it was developed. Notably it also offers the ability to impose one of three different brake curves on a train-by-train basis with only one set of balises.

Punktförmige Zugbeeinflussung - Wikipedia

en.wikipedia.org

It does help that DB uses speed signalling. Rather more complex to come up with something like that for block signalling.

 

[lammergeier]

That's a fair point. I'm eagerly awaiting the publication of the report into the LUMO incident - and whether there's any holes in the TPWS safety net the train slipped through that can be fixed without going for ATP.

Flashing yellows being the obvious major hole in the TPWS net.

 

[MarkyT]

It does help that DB uses speed signalling. Rather more complex to come up with something like that for block signalling.

Doesn't really make much of a difference to warning and protection systems. The interlocking knows which route is set from which a speed can always be derived, regardless of how that is conveyed to the driver by aspect, route or speed indication. The bigger problem with AWS is closely following trains routinely going near full line speed on double yellows that step up to green on approach, particularly on the busy Southern Region which is partly why management there was resistant to implementing the system. The fear is that repeated routine cancelling of warnings can become habitual and then a 'real' warning risks being ignored.

== Doublepost prevention - post automatically merged: 1 May 2023 ==

Flashing yellows being the obvious major hole in the TPWS net.

And where other forms of junction signalling arrangements not approach controlled from red are deployed for divergences, such as splitting distant and free yellow. While, as I said earlier, an approach release from red scenario can theoretically be partly protected using normal OSS functionality, all such signals with divergences at significantly lower speed are not equipped with TPWS as they don't all also have convergence/crossing type of conflicts in route to protect, which is the usual criterion for fitment.

 

[lammergeier]

]

And where other forms of junction signalling arrangements not approach controlled from red are deployed for divergences, such as splitting distant and free yellow. While, as I said earlier, an approach release from red scenario can theoretically be partly protected using normal OSS functionality, all such signals with divergences at significantly lower speed are not equipped with TPWS as they don't all also have convergence/crossing type of conflicts in route to protect, which is the usual criterion for fitment.

I would suggest where diverging routes are concerned, the potential risk is derailment due to significant overspeed as opposed to collision mitigation. Fairly easy for a driver to cancel AWS warnings but not react.

Where significant reductions in speed occur on plain line they are often protected by TPWS irrespective of signal aspects, but this protection is not provided at diverging junctions.

 

[Railsigns]

And where other forms of junction signalling arrangements not approach controlled from red are deployed for divergences, such as splitting distant and free yellow.

Reductions in permissible speed through diverging routes with splitting distant signals on approach are assessed for TPWS fitment, subject to the usual fitment rules that take into account the speeds involved.

 

[Irascible]

Yes, PZB (as Indusi was renamed) is a remarkable achievement in and of itself, especially considering how long ago it was developed. Notably it also offers the ability to impose one of three different brake curves on a train-by-train basis with only one set of balises.

If I'm remembering it correctly, TVM just sends "aspect codes" ( my phrase ) to the train & it's up to the onboard system to show what they actually mean ( and to decide if the train's within the braking curve ) - so a TGV and a 92 can receive the same code from a balise but show a different speed limit in the cab as they might have different tables onboard ( I don't know if they actually *do* - one would think a 92 doesn't have a cleared-for-300km/h "aspect!" at least- but it's possible ). So, I guess if there was a use you could also signal a more conventional line with TVM & have greater granularity in speed control by using a different table just for that route, perhaps locked in by a token sent to the train. There's not been any apparent need to develop it though.

I like PZB even if I've forgotten a lot about it. It's taken quite some development though.
[edited for tablet going nuts]

 

[MarkyT]

If I'm remembering it correctly, TVM just sends "aspect codes" ( my phrase ) to the train & it's up to the onboard system to show what they actually mean ( and to decide if the train's within the braking curve ) - so a TGV and a 92 can receive the same code from a balise but show a different speed limit in the cab as they might have different tables onboard ( I don't know if they actually *do* - one would think a 92 doesn't have a cleared-for-300km/h "aspect!" at least- but it's possible ). So, I guess if there was a use you could also signal a more conventional line with TVM & have greater granularity in speed control by using a different table just for that route, perhaps locked in by a token sent to the trian. There's not been any apparent need to develop it though.

U kuje OZV eveb if I've forgotten a lot about it. It's taken quite some development though.

Rather than the digital balises of systems like ETCS, TVM430 and the previous TVM300 rely primarily on audio frequency track circuits on which are modulated a range of control code frequencies from which the train can calculate its safe speed envelope. French trains with TVM typically also have the 'limited supervision' KVB system onboard for use on SNCF classic lines, at speeds lower than TVM is capable of supervising. KVB uses active switched digital balises very similar to those of ETCS. TVM specs also include another intermittent inductive loop system with beacons between the rails for certain additional system commands on high speed lines such as arming/disarming the TVM, switching power systems, raising/lowering pantographs and opening/closing ventilation inlets on the train at tunnels.

 

[Irascible]

Thanks for the correction / elaboration - it's been a while.I had a ferling balise wasn't the right word, but it was functionally good enough

 

[MarkyT]

I would suggest where diverging routes are concerned, the potential risk is derailment due to significant overspeed as opposed to collision mitigation. Fairly easy for a driver to cancel AWS warnings but not react.

Junction overspeed is the major risk being discussed here. Although on the latest trains that emulate AWS in their ETCS equipment, it might be fairly easy to implement additional controls to provide some ongoing restriction for a time or distance after a caution indication, as in PZB, the operational necessity to drive on continuous double yellows and the placing of the magnet ~200m on approach to the signal in UK rather than at the post in Germany makes this more difficult.

Where significant reductions in speed occur on plain line they are often protected by TPWS irrespective of signal aspects, but this protection is not provided at diverging junctions.

As I said, there is some protection in cases where approach release from red applies (AR), AND the signal already has TPWS for other reasons. In many cases with good visibility the signal and JI will have cleared well before the train has reached the OSS though, and even if the train successfully passes the active OSS at under the trap speed, there's nothing to prevent a driver from accelerating again, and, if misinterpreting the route indication, potentially speeding significantly through the junction turnout(s). This was the Bletchley light engine scenario that led to a junction overspeed derailment in the TPWS era.

Reductions in permissible speed through diverging routes with splitting distant signals on approach are assessed for TPWS fitment, subject to the usual fitment rules that take into account the speeds involved.

Additional PSR-style OSS grids, but only switched on when the diverging lower speed route is set? If so, it seems there is a methodology that might also be applied in other non-AR cases. Possibly academic for the northern approaches to Peterborough in the LUMO incident though, as the area is due to get an ETCS overlay which should provide much better full ATP supervision for suitably equipped trains.

 

[Railsigns]

Additional PSR-style OSS grids, but only switched on when the diverging lower speed route is set?

Yes, but with the OSS loops conditioned by the lie of the facing points, such as fitted on approach to Bicester South Junction (100 mph approach on the Down Main onto either of two 40 mph divergences).

 

[lammergeier]

.

As I said, there is some protection in cases where approach release from red applies (AR), AND the signal already has TPWS for other reasons. In many cases with good visibility the signal and JI will have cleared well before the train has reached the OSS though, and even if the train successfully passes the active OSS at under the trap speed, there's nothing to prevent a driver from accelerating again, and, if misinterpreting the route indication, potentially speeding significantly through the junction turnout(s). This was the Bletchley light engine scenario that led to a junction overspeed derailment in the TPWS era.
.

At flashing yellows, however, which is the blind spot I was talking about, there is zero(?) TPWS protection. Approach Release is a separate thing entirely and often the flank protection provided means there is less risk than the risk of derailment which can come with flashing yellows and no TPWS protection.

 

[MarkyT]

At flashing yellows, however, which is the blind spot I was talking about, there is zero(?) TPWS protection. Approach Release is a separate thing entirely and often the flank protection provided means there is less risk than the risk of derailment which can come with flashing yellows and no TPWS protection.

In a standard flashing yellow sequence, all signals passed are showing proceed so no standard signal-related TPWS transponder will be energised at any time and thus no protection can be provided, as you say. @Railsigns explained there is a method for fitting a speed control OSS for a non-approach released from red divergence, as deployed fairly recently at Bicester with splitting distants. The same method might plausibly be used for a flashing yellow application if desired. Switching PSR-style OSS loops using point detection seems a fairly simple technique, although it might become more difficult in a complex junction.

 

[edwin_m]

In a standard flashing yellow sequence, all signals passed are showing proceed so no standard signal-related TPWS transponder will be energised at any time and thus no protection can be provided, as you say. @Railsigns explained there is a method for fitting a speed control OSS for a non-approach released from red divergence, as deployed fairly recently at Bicester with splitting distants. The same method might plausibly be used for a flashing yellow application if desired. Switching PSR-style OSS loops using point detection seems a fairly simple technique, although it might become more difficult in a complex junction.

Rather than a long tail cable all the way from the points to the loops, I think this would be done by adding some extra controls from the interlocking. Reasonably straightforward on a CBI if there are spare trackside outputs available.

 

[MarkyT]

Rather than a long tail cable all the way from the points to the loops, I think this would be done by adding some extra controls from the interlocking. Reasonably straightforward on a CBI if there are spare trackside outputs available.

Agreed, especially in the case of a large junction, where the selection control logic might be quite complex and more easily accomplished in a line of data.

 

[HSTEd]

Somewhat related to the topic at hand, as it is likely to take decades to resignal the network even once ETCS is adopted (with ATP built in obviously), that it might be worth fitting ETCS LEvel 1LS on the rest of the network?

That would allow all the AWS and TPWS kit to be dispensed with and the training requirements abolished.

 

[zwk500]

Somewhat related to the topic at hand, as it is likely to take decades to resignal the network even once ETCS is adopted (with ATP built in obviously), that it might be worth fitting ETCS LEvel 1LS on the rest of the network?

That would allow all the AWS and TPWS kit to be dispensed with and the training requirements abolished.

It's not worth the cost. The benefits of L1 LS over AWS/TPWS is small and the cost is big.

 

[MarkyT]

Somewhat related to the topic at hand, as it is likely to take decades to resignal the network even once ETCS is adopted (with ATP built in obviously), that it might be worth fitting ETCS LEvel 1LS on the rest of the network?

That would allow all the AWS and TPWS kit to be dispensed with and the training requirements abolished.

That's effectively what Switzerland developed, re-engineering their legacy protection systems initially to use ETCS hardware as a stepping stone to, first, standard L1LS, then, higher level ETCS functionality as required. In parallel, they also installed standard Level 2 on their newer higher speed and some of their busiest trunk lines. Thus they became the first European country to offer full ETCS interoperability throughout their entire standard gauge network in 2017. Belgium is engaged in implementing something similar. France were pioneers with their KVB system from the 1990s, using digital balises very similar to active Eurobalises, using the same carrier frequency I think and thus potentially readable by standard onboard ETCS balise reading hardware if a train has the correct emulation software. Switzerland persuaded Germany and France to co-sponsor inclusion of standard LS (Limited Supervision) functionality into the later baselines for ETCS. ISTR reading Germany is considering converting their PZB system to use ETCS hardware.

 

[HSTEd]

It's not worth the cost. The benefits of L1 LS over AWS/TPWS is small and the cost is big.

If its a choice between maintaining the complete AWS/TPWS system vs installing the complete ETCS L1LS system, sure.

But that is not what the situation will be, the question will be installing trackside ETCS L1LS equipment out of standard parts, vs maintaining the complete AWS/TPWS system.

If ETCS Level 1LS is adopted, all the relevant AWS/TPWS training overhead and all the supply chain stuff vanishes.
A lot of expensive to validate and maintain train-side equipment just goes in the bin without replacement.

The march of ETCS Level 2 in new installations will probably do a real number on the commercial viability of the supply chain for AWS/TPWS parts.

 

[zwk500]

If its a choice between maintaining the complete AWS/TPWS system vs installing the complete ETCS L1LS system, sure.

But that is not what the situation will be, the question will be installing trackside ETCS L1LS equipment out of standard parts, vs maintaining the complete AWS/TPWS system.

The issue for ETCS L1 is you still need to do the legwork in the background, including fitting all trains with DMIs, doing all the data work, converting interlockings to RBCs, and getting all the speed profiles etc. You are saving minimal amounts of money on only fitting balises at important signals not every signal, and not getting any of the capacity benefits alongside it.

If ETCS Level 1LS is adopted, all the relevant AWS/TPWS training overhead and all the supply chain stuff vanishes.
A lot of expensive to validate and maintain train-side equipment just goes in the bin without replacement.

The march of ETCS Level 2 in new installations will probably do a real number on the commercial viability of the supply chain for AWS/TPWS parts.

I don't think ETCS L2 will roll out quickly enough to reduce AWS/TPWS to an unviable number of installations that makes the intermediate step of ETCS L1-LS worthwhile.

 

[Annetts key]

Perhaps you could tell us when the last fatal accident was that could have been prevented by ATP but AWS/TPWS would have been/was ineffective?

I don’t know the answer to that. But I do know of some of the holes in AWS and TPWS.

Not all signals are fitted with TPWS. Indeed, there are as many if not significantly more signals that are not fitted with TPWS compared to those that are fitted. This is partly why TPWS costs less compared to a “standard” ATP system (where all main signals are fitted).

Hence if a train stops for any reason in a section that is protected by a signal that is not fitted with TPWS (such as the vast majority of signals on plain line), then on a non-ATP fitted line, or with a non-ATP fitted train, apart from the driver of the next train obeying the signal aspects, there is only AWS that can intervene (assuming that the driver does not acknowledge it and continues on).

Secondly, TPWS is only considered effective for speeds up to 75MPH. TPWS+ increases this to 100MPH (105MPH for some trains). For higher speed lines, “double blocking” (two consecutive red signals, both of which are fitted with TPWS) are one option.

TPWS+ and “double blocking” of course increases the system cost significantly.

TPWS+ requires not only extra loops and associated TPWS equipment, but also normally extra location cases/cupboards and often extra lineside cables to the location cases/cupboards where the signal control circuitry is located.

And all TPWS and TPWS+ installations require changes to the interlocking (GWML ATP for relay based interlocking does not need changes, as it can use APLITS instead).

GWML ATP of course can cope with all speeds used on the lines it’s fitted to, including 125MPH lines with no extra equipment other than the normal ATP installation.

TPWS can only spot supervise speed restrictions where it has been specifically provided. Again, just like TPWS+, extra equipment is required for each and every speed restriction.

Again, this is standard on ATP, with normally no extra equipment being needed (occasionally if there are multiple speed and/or junction speed limits, an extra beacon may be needed). Plus ATP provides continuous speed limit supervision, not just spot supervision.

And as hinted at above, GWML ATP does supervise junction speed limits.

I agree, and have to add that IMHO TPWS is one of the greatest single improvements in rail safety since interlocking of signalling.

Totally disagree. I would say that route relay interlocking is the one thing that brought a lot of technologies together to form a very safe system, that virtually eliminates incidents caused by signaller error (during normal operation) and which helped to reduce the accident rate substantially. Of course the difficulty is that you can’t prove a negative. Because route relay interlocking prevents accidents and incidents, it’s impossible to know how many have been prevented…

Apart from the limitations of TPWS that I listed above, the second problem with TPWS, is that the location case/cupboard/relay room equipment is not actually very reliable.

Hence there is an almost continuous stream of TPWS faults reported by signallers every week. Compared to the the location case/cupboard and trackside equipment for GWML ATP where faults are better described as occasional. Although I can’t comment on what the situation is with cab/train faults.

I'd put the track circuit above TPWS, personally. TPWS is great for what it is, but it does leave some important problems - it cannot cope with different speeds on diverging routes, it cannot adjust to different speed profiles, and it provides no advance warning before tripping the brakes. I also have it in the back of my mind that it's not fail-safe, as the grids require power to activate.

TPWS is a safety system but is not fail safe. There is no way for the equipment on the train to know where the TPWS loops are located, hence if for any reason, either or both of the loops do not transmit their fixed frequency, the on train equipment will not trigger the brakes or generate any error indication.

ATP is a considered a fail safe system (although it’s unlikely it would meet modern ideas on this). The reason being that the on-train equipment does know where the next ATP beacon is supposed to be located. If within a set tolerance, the on-train equipment does not receive a valid ATP message, the driver will be alerted via a five second brake application and an error code. The system will then go from full supervision to partial supervision.

In principle it ought to be possible to trigger different TPWS loops at the approach to a junction depending on the route set. This information is readily available in the interlocking but would need to be passed to the loops via a new control to the trackside. I assume this didn't happen at Peterborough because of the difficulty in modifying a 1970s-era relay interlocking, but does anyone know if it is done on newer signalling schemes elsewhere?

I don’t recall seeing this. There are AWS that become active for lower speed junctions on the approach to the junction.

I was involved in the early days of TPWS circa 1994, when some colleagues analysed the accident reports back to around 1968 to assess whether the accident would have been less severe, or would not have happened at all, if a TPWS-like system had been fitted at the time. This also allowed for the fact that the older rolling stock involved in many of these accidents would now be replaced by newer and more survivable designs. The conclusion was that TPWS would avert about 70% of the casualties that an ATP system would, which was noted with some skepticism by Uff-Cullen but borne out some years later in an analysis by Stanley Hall in Modern Railways magazine. TPWS+ has probably improved this further since then. Even then the predicted cost was near the maximum that the expected casualty reduction would justify, and the cost later increased due to measures such as proving the system operational.

But that ignores where we could have been today if ATP had been a requirement for all new rolling stock and all new signalling schemes (or relock schemes, or substantial alterations) from say 1990 onwards.

The lack of failsafe was known and agreed at the time by senior people in BR and Railtrack (who unlike some of their successors actually had both technical knowledge and influence within the organisation). To cause a SPAD-related accident it is necessary for the TPWS to fail but also for some other event to occur such as the driver misinterpreting the signal or braking too late. Both of these should be very unusual circumstances so the chance of both at the same time is infinitesimal - provided TPWS failures are promptly identified and remedied so they don't remain as latent faults. It's worth noting that a modern safety integrity analysis wouldn't consider AWS to be failsafe either.

and

It is not fail-safe, but it is protected somewhat by the fact that a power failure, or the majority of other fault conditions, will lead to the proving contacts for the TPWS circuitry not making. This will alert the signaller of the failure and prevent the signal in rear from clearing whilst the signal associated with the defective TPWS is at danger. It doesn't make it impossible for all the holes in the analogical Swiss cheese to line up, but it does add a few more layers to it.

The trouble is, with the number of TPWS failures that occur, I would say that the risk is not infinitesimal. Especially as the current climate in busy locations/lines is to keep trains running until a suitable line block can be arranged later during the night.

Okay, if the signal that is protecting the signal that has a failed TPWS also, itself is fitted with TPWS, then the risk may be considered reasonable. But not all are (sorry, I can’t provide any numbers).

Interesting, could you expand on why? I thought the magnet being suppressed with an electromagnet was a fail-safe design.

The big problem with AWS has and still is, the on train equipment. The track mounted equipment (and the control equipment) is considered fail-safe. If power is lost, or a “right side” failure occurs (where suppressed permanent magnets are used), the permanent magnet will still “generate” a magnetic field that the on-train equipment can detect.

Unfortunately, it appears that the on-train equipment is not fail safe. But you would have to get input from a train fitter or other suitable person who knows about the on-train equipment.

In terms of reliability, the track mounted equipment (and the control equipment) is generally reliable. The modern suppressed permanent magnets appear at the moment to be the most unreliable part (although they fail safe side).

The biggest problem with AWS failures, is that many drivers don’t report the problem. I’ve experienced failures (that cause the electro-magnet to not be energised, hence the driver gets a horn instead of a bell, i.e. code 2 fault) where one train reports the problem, but none of the other drivers reports the problem. Yet it’s obvious when the signal technician locates the problem, that it must have been faulty for hours, possibly days…

Now, before anyone piles in to make the case for TPWS, or to explain why ATP was not considered cost effective, may I point out that all this has already been discussed to death in another thread. I don’t think it’s worthwhile going over this again.

However, I am very disappointed that ERTMS/ETCS has not been brought in as originally intended to cover our high speed lines (where TPWS is not as effective). On the GWML, the existing ATP was supposed to have been replaced with ERTMS/ETCS by now according to one briefing that I received many, many years ago. And the GWML was apparently not first in the list…

So, from my point of view, what we should be discussing is why it’s taking so long for our railways to introduce ERTMS/ETCS?

I understand that level 1 or level 2 (mixed operation/overlay) is considered not to be value for money. But, again, if from say a couple of years after the Cambrian line was deemed fully operational, where could we have been today if it had been a requirement for all new rolling stock and all new signalling schemes (or relock schemes, or substantial alterations) to be fitted?

 

[zwk500]

First off, thank you so much for all that information.

So, from my point of view, what we should be discussing is why it’s taking so long for our railways to introduce ERTMS/ETCS?

Finance, is the primary reason. But also questionable record keeping and problematic information sharing of data already gathered hindering review and design exercises.

I understand that level 1 or level 2 (mixed operation/overlay) is considered not to be value for money. But, again, if from say a couple of years after the Cambrian line was deemed fully operational, where could we have been today if it had been a requirement for all new rolling stock and all new signalling schemes (or relock schemes, or substantial alterations) to be fitted?

AIUI the equipment fitted to the Cambrian is not compatible with the latest versions of ETCS standards and requirements. You have to look to the treasury for the decision to spec trains to be ETCS-ready not ETCS-equipped until the IET programme came in.

 

[HSTEd]

The issue for ETCS L1 is you still need to do the legwork in the background, including fitting all trains with DMIs

Almost all trains are going to have to be fitted with DMIs fairly soon anyway.

It won't take many main line ETCS resignalling projects until almost all rolling stock is going to have to be fitted regardless.

With ECML-South and MML alone you probably need to fit a substantial fraction of the entire rolling stock fleet. And isn't most of the new fleet fitted from new anyway?

, doing all the data work, converting interlockings to RBCs

L1LS doesn't necessarily require any interlocking side changes.
The Swiss installations are/were drop in trackside replacements for Signum and ZUB.

You don't even need fundamental speed profile information if you don't want, it can just be used in the TPWS style to transmit a overspeed limit and/or "stop now" information.

I don't think ETCS L2 will roll out quickly enough to reduce AWS/TPWS to an unviable number of installations that makes the intermediate step of ETCS L1-LS worthwhile.

Well if it doesn't roll out fairly rapidly the entire railway may just collapse due to the inability to resignal fast enough!
If ETCS on the southern ECML delivers even a fraction of what it supposedly can, I don't think there will be many more, if any, conventional resignalling projects.

At which point there won't look to be much of a future in AWS or TPWS.

 

[Railsigns]

Secondly, TPWS is only considered effective for speeds up to 75MPH. TPWS+ increases this to 100MPH (105MPH for some trains). For higher speed lines, “double blocking” (two consecutive red signals, both of which are fitted with TPWS) are one option.

Those 75 mph / 100 mph theoretical limits have no relevance within the present-day TPWS design process, which uses a TPWS placer tool to calculate an effective design tailored to each individual signal.

One of the new signals at Carstairs for instance is approached at 125 mph and has effective TPWS comprising two OSS fitments, with no double-blocking or TPWS fitment at the signal in rear. With one OSS at 375 m from the signal and the other at 825 m, this is an example of what used to be termed TPWS+.

 

[Annetts key]

When I was first briefed about TPWS, it was implied that this system was a stop-gap…

Yes, the Cambrian ETCS is not compatible with the latest versions of ETCS standards. But that’s not surprising as technology moves forward.

Similarly, there are other signalling technologies which are obsolete, hence Network Rail have had to replace them.

== Doublepost prevention - post automatically merged: 3 May 2023 ==

Those 75 mph / 100 mph theoretical limits have no relevance within the present-day TPWS design process, which uses a TPWS placer tool to calculate an effective design tailored to each individual signal.

One of the new signals at Carstairs for instance is approached at 125 mph and has effective TPWS comprising two OSS fitments, with no double-blocking or TPWS fitment at the signal in rear. With one OSS at 375 m from the signal and the other at 825 m, this is an example of what used to be termed TPWS+.

So how is the design done to overcome the fundamental compromise of the different braking curves of different rolling stock and of the distance between signals? And that of a driver going slowly enough not to trip it at the (first) OSS+, but still too fast for the OSS to be effective?

I know there was talk at one time of having two sets of OSS+ loops (or is that what you are talking about).

 

[MarkyT]

L1LS doesn't necessarily require any interlocking side changes.
The Swiss installations are/were drop in trackside replacements for Signum and ZUB.

You don't even need fundamental speed profile information if you don't want, it can just be used in the TPWS style to transmit a overspeed limit and/or "stop now" information.

Swiss EuroZub and EuroSignum didn't require a full ETCS fitment onboard the train including the DMI. Similar to the TPWS approach, a 'drop in' cut-down box of electronic wizardry was invented to replace the legacy system onboard in a deliberate attempt to make retrofits to older traction practical and not require great changes to rules and procedures. Trains so equipped couldn't go onto the L2 lines clearly, but once EuroZub/Signum was migrated quickly to the new standard L1 LS, even foreign modern traction with full later baseline ETCS became natively compatible with the entire Swiss standard gauge network, at least from a signalling point of view.

 

[Railsigns]

So how is the design done to overcome the fundamental compromise of the different braking curves of different rolling stock and of the distance between signals? And that of a driver going slowly enough not to trip it at the (first) OSS+, but still too fast for the OSS to be effective?

The TPWS effectiveness calculations give values for different braking rates.

TPWS design always considers that trains may approach the signal at any speed up to the permissible speed. This means that a train travelling too slowly to be triggered at the first OSS will be tripped either at a subsequent OSS or at the TSS and still be brought to stand before the conflict point. If TPWS designs only catered for trains approaching at full permissible speed, there'd be no point in ever having more than one OSS.

 

[MarkyT]

Yes, the Cambrian ETCS is not compatible with the latest versions of ETCS standards. But that’s not surprising as technology moves forward.

It's being upgraded to latest baseline. I don't know the progress status of this project. I think the work is needed for compatibility with the new TfW fleet, some of which will come already ETCS equipped for the line, and to address shortcomings with the original application design such as the little problem that occurred regarding temporary speed restrictions.

 

[Falcon1200]

Totally disagree.

Fair enough, you have your opinion and I have mine. But I would point to the virtual elimination of collisions caused by SPADs as a direct benefit of TPWS, having been involved in a minor way with such incidents as Bellgrove and Newton, both of which TPWS would have prevented. Not that the railway should become complacent of course, or cease improving.

 

[zwk500]

Almost all trains are going to have to be fitted with DMIs fairly soon anyway.

It won't take many main line ETCS resignalling projects until almost all rolling stock is going to have to be fitted regardless.

This is true, but is a reason to go to L2 and take advantage of the functionality.

With ECML-South and MML alone you probably need to fit a substantial fraction of the entire rolling stock fleet. And isn't most of the new fleet fitted from new anyway?

There's a hell of a lot of trains in the country that don't run on those lines, but yes the new fleet will be ETCS ready anyway.

L1LS doesn't necessarily require any interlocking side changes.
The Swiss installations are/were drop in trackside replacements for Signum and ZUB.

My understanding is that the swiss systems had been migrating to align with ETCS anyway, so were close enough to have a conversion system. AWS/TPWS are nowhere near the ETCS architecture.

You don't even need fundamental speed profile information if you don't want, it can just be used in the TPWS style to transmit a overspeed limit and/or "stop now" information.

If you're doing that it seems a poor return on the investment.

Well if it doesn't roll out fairly rapidly the entire railway may just collapse due to the inability to resignal fast enough!

The railway has been using different signalling systems for centuries. The idea that the railway would collapse because it can't transition to ETCS fast enough is laughable. We're still using Electric Token Block or Train Staff working on some line. We'd certainly be able to keep the AWS kit running on a handful of installations. RETB never got beyond 4 places.

If ETCS on the southern ECML delivers even a fraction of what it supposedly can, I don't think there will be many more, if any, conventional resignalling projects.

It will certainly be the intent to avoid it but we will have legacy signalling systems for a long time yet.

At which point there won't look to be much of a future in AWS or TPWS.

AIUI the plan to migrate everything into the ROCs had a final date of 2050 for the last closure when it was originally announced in 2015, and we're already quite far behind that proposed timescale. ETCS rollout will speed up as it becomes more common over the country and training and procedures are tightened up. But we have 20,000 track miles in the country and currently use a vast number of systems many of which are still Victorian in principle, not to mention the differences in underlying technologies in TCB systems. I would not be so quick to say AWS/TPWS has its days numbered.

 

[Annetts key]

Fair enough, you have your opinion and I have mine. But I would point to the virtual elimination of collisions caused by SPADs as a direct benefit of TPWS, having been involved in a minor way with such incidents as Bellgrove and Newton, both of which TPWS would have prevented. Not that the railway should become complacent of course, or cease improving.

The primary reason that I don’t consider TPWS to be one of the greatest single improvements in rail safety since interlocking of signalling, is because it was one element of a number of things.

1. The dangers of SPADs were already known long before the Ladbroke Grove disaster.
2. A system to protect against them already existed (ATP).
3. The very high danger of particular junction layouts if a train passed certain signal(s) at red were already known about, but saving money was considered above having a better junction layout (single lead junctions instead of double lead junctions for example).
4. As well as the introduction of TPWS, various other measures were introduced or other changes made.
   • Such as better and more serious investigations into why they occurred rather than just blaming the driver.
   • Signal sighting being improved for some problem signals.
   • Human factors are now taken seriously - especially important for train drivers, where their concentration is important (side note, I wish better consideration would be applied during equipment design for signallers, platform staff and signal technicians).
   • Provision of banner signals or adding extra ‘distant’ colour light signals to reduce the distance between the driver seeing a caution (yellow) aspect and the junction signal at red (yes, sounds backwards logic, but this has been done where previously the distance between signals far exceeded the service braking distance).
   • In new signalling designs, possible provision of double blocking for potentially high risk areas, or other mitigation measures.
   • Countdown or triangular “signal xxx yards ahead” signs.
   • Wider adoption of defensive driving.
   • Publication of the signals that had the highest number of SPADs in the WON.
   The above list is not exhaustive as there may be other things that I have forgotten.

On points 1 and 3, one of my managers (engineers) and myself had a long discussion about this some years before the Ladbroke Grove disaster. I can’t remember when (but I think it was late 1980s or very early 1990s), but a serous SPAD (thankfully no collision) elsewhere in southern England at a single lead junction occurred and hence they were looking at the single lead junctions on our area. And if any reasonable cost improvements could be made.

Also, just to be clear, I do agree that TPWS is an improvement and has definitely improved safety on the railways. But it was and is only part of what was done.

Just a further note. The reason that ATP was not rolled out was cost. But a cut down ATP system was possible (at least in principle for the type used on the GWML). Not all signals had to be fitted and the operation of it could have been changed for lines that were only partially fitted. Indeed, not all signals were originally fitted for the GWML pilot ATP system, deliberate gaps were left (the gaps have since been removed by fitting the relevant signals with ATP).