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

[bramling]

Equally, had the view been taken in the early 2000's to focus on the ATP rollout because it provides a higher level of protection than TPWS can, we'd have prevented multiple overspeed incidents.

But so far the number of fatalities that 100% ATP coverage would have prevented appears to be 1. And note that we still managed to have a (fatal) head-on collision on one of the few sections of mainline railway which *is* fitted with continuous ATP. This all assumes that ATP could have been rolled out system-wide in that time anyway, which is by no means a given.

It is a fair point that any number of these overspeeds or whatever could have led to a more serious outcome - indeed we are probably very lucky at least one if not more didn’t. But here we are, money is finite…

 

[Sorcerer]

ETCS should make TPWS unnecessary as it constantly supervises unlike AWS and TPWS being intermittent.

ERTMS/ETCS would make TPWS redundant as it covers the functionality of TPWS in a better way. The planned digital signalling rollout cannot possibly cover the entire country in one go (resources or money), nor would it be wise to do so.

Worth pointing out that there's ETCS Level 1, which is essentially just an ATP layer on conventional signalling, and ETCS level 2, which is the in-cab bit with Block Markers.
Level 1 would make TPWS redundant in covered areas, but Network Rail seem to try to be avoiding Level 1 as much as possible, presumably because it incurs significant cost in converting the interlocking but doesn't provide meaningful capacity upgrades.
Level 2 can be 'overlaid' on existing signalling (as is done on the GWML and Thameslink Core), which is usually provided when there's no plans to convert all trains to ETCS, or as an interim step to signals-away. In a L2 overlay situation, TPWS would still be required for trains not fitted with ETCS.
In a proper Level 2 situation, TPWS would be redundant (e.g. Moorgate-Drayton Park, although I've got no idea if they did recover the grids there.)

If you google 'Enhanced TPWS' you should be able to find various reports comparing the collision risk of ETCS, GW-ATP, normal TPWS and 'Enhanced TPWS' which is where it's fitted to most/all signals and potentially adding more loops at high-risk locations.

Interesting. Ideally then I would like to say that the focus should be on ETCS rollout on the network, though at the same time this is an answer I have to a question that is based on an incident in which the RAIB haven't yet determined the cause of yet. I think that I'll probably be more confident once they report their findings and subsequent recommendations.

 

[Bald Rick]

I think I’m right in saying this is the first actual accident where non-fitment of TPWS to certain signals has been a factor? Can anyone point to any others, or any known near-misses?

Depends on the definition of ‘accident’, but AIUI this is the first major incident involving passenger injuries due to a TPWS preventable SPAD (if that is what it was) for a long time.

Railway safety is a science of risk, probability and finance. There are far bigger risk precursors that cause many more fatalties and injuries that could be prioritised ahead of TPWS roll out to every signal.

 

[bramling]

Interesting. Ideally then I would like to say that the focus should be on ETCS rollout on the network, though at the same time this is an answer I have to a question that is based on an incident in which the RAIB haven't yet determined the cause of yet. I think that I'll probably be more confident once they report their findings and subsequent recommendations.

I don’t think it’s unreasonable to have already reached a conclusion that Friday’s accident would very likely have been prevented by TPWS, and even if not prevented that certainly very significantly reduced in severity.

Where one goes with that conclusion is another matter entirely…

 

[Nicholas Lewis]

That’s the risk which you crossover. You focus attention on the crossover speed, forget the previous signal was a single yellow and just power up once you know you’ve cleared the crossovers. That’s why TPWS should be installed at signals following a crossover.

Add to that, he’d probably done that move so many times, but never on a single yellow. Normally the fast trains would be well clear of the sections ahead.

What is the normal aspect sequence for a train leaving platform 1 and crossing over to the Up Fast running under effectively clear signals?

Given the divergence speed is 50mph for a US-EF move WH356 will have approach control but it only has one divergent route and divergent speed differential is 40mph so i'm not sure that necessitates approach control from Red. So it was more than likely single/double yellow with position 4 junction indicator so subconsciously used to receiving horn warnings in this area. Anyhow RAIB report will disclose all plus what the train warnings being received were.

 

[bramling]

Depends on the definition of ‘accident’, but AIUI this is the first major incident involving passenger injuries due to a TPWS preventable SPAD (if that is what it was) for a long time.

That’s what I was thinking. I cannot readily think of any other serious accident that was TPWS-preventable, since Ladbroke Grove.

I’m just having a look at Wikipedia (realise this isn’t the best source!) which does give a list of “notable” rail accidents this century, and the general theme seems to be either collisions with road vehicles or infrastructure issues.

One thing worth noting IMO is that if one decides to include the Croydon tram derailment, things are skewed a bit. However this is only loosely connected, as whilst it involves train protection in the general sense, it isn’t relevant to TPWS.

Railway safety is a science of risk, probability and finance. There are far bigger risk precursors that cause many more fatalties and injuries that could be prioritised ahead of TPWS roll out to every signal.

Can’t disagree with this. It may well be that one fatality indicates that the decision to only fit TPWS to selected signals was actually a good decision. Suspect it will be a focus of some discussion over the next few months though.

 

[HSTEd]

The RSSB have looked at it: https://www.rssb.co.uk/research-catalogue/CatalogueItem/T1043

Not sure if the documents are available to non-registered people but the summary should be visible.

I'm not sure why would you pay for ETCS L1LS to be installed and then maintain the existing train protection systems?

Surely it would be better to accelerate ETCS fitment than maintain multiple train protection systems?

What fraction of the rolling stock fleet is not ETCS ready and is not expected to remain in service long enough to need ETCS fitment regardless?

EDIT:
Especially as that report is from ten years ago, ETCS fitment has been ongoing that entire time.

I was proposing the wholesale replacement of AWS and TPWS with ETCS along the lines of the replacement on the Swiss railway system.

 

[Sorcerer]

I don’t think it’s unreasonable to have already reached a conclusion that Friday’s accident would very likely have been prevented by TPWS, and even if not prevented that certainly very significantly reduced in severity.

Where one goes with that conclusion is another matter entirely…

Truthfully I haven't looked into it much, so I didn't want to draw conclusions so early, but I am quite confident saying any future fitting to UK signals should be some form of ETCS since it's a very well proven system already.

 

[edwin_m]

Another thought to consider is that the types of accident preventable by TPWS have also been made less severe if they do happen, by measures such as deformable structures and interiors that reduce secondary casualties. The latter also helps mitigate the effect of accidents from other causes. So even without TPWS or ATP, the amount of casualties arising from SPADs would most likely have reduced over time.

Imagine the consequences of a 45mph impact as reported for Bedford, had it been Mk1 or older stock.

 

[Snow1964]

Indeed. But also it is important to wait until the RAIB have thoroughly investigated to understand whether this accident requires any changes to the risk calculations, and if so what they should be.

Exactly, RAIB should determine risks and recommendations

It looks horribly like AWS led to sequence of events happening, had it not broke on the 810, wouldn't have been sitting there on the fast line.

AWS is a system that has been around from 1950s, better than nothing, but not perfect, and in a way, it's out of date but still in use.

I am sure RAIB will look at the AWS failure on a new train (although I do not know if the AWS equipment was brand new, or secondhand from a withdrawn unit). It's upto RAIB to decide if fitting an old system to a new train changes risk, not for me to speculate on.

 

[zwk500]

I'm not sure why would you pay for ETCS L1LS to be installed and then maintain the existing train protection systems?

Surely it would be better to accelerate ETCS fitment than maintain multiple train protection systems?

What fraction of the rolling stock fleet is not ETCS ready and is not expected to remain in service long enough to need ETCS fitment regardless?

EDIT:
Especially as that report is from ten years ago, ETCS fitment has been ongoing that entire time.

I was proposing the wholesale replacement of AWS and TPWS with ETCS along the lines of the replacement on the Swiss railway system.

You will need to maintain legacy systems until all trains are ETCS fitted (and staff trained, etc)

AIUI the Swiss spent quite a long time harmonising their national system to be very close to ETCS L1 prior to beginning the conversion system.
AWS and TPWS are not compatible with ETCS in the same way.
One of the papers I read did dicuss a TPWS-E option where the TPWS functionality was maintained but the grids were replaced with eurobalises. That would presumably still have counted as Level NTC rather than L1 though.

 

[HSTEd]

You will need to maintain legacy systems until all trains are ETCS fitted (and staff trained, etc)

Well...how many trains are left that aren't ETCS ready? Didn't the decision to start preparing new stock for that happen nearly 20 years ago now?

I'd think you'd prioritise fitment of trains over the infrastructure changes, so that the bulk of the track never actually has to be dual fitted.

Anything on the infrastructure side seems to be far more expensive than the rolling stock side these days.

 

[Bald Rick]

Anything on the infrastructure side seems to be far more expensive than the rolling stock side these days.

The infrastructure is cheap.

The cost is partly in rolling stock fitment (there is lots not ETCS ready, and even that which is is still expensive), and partly in all the business change activity that comes from it. Preparing documentstion for the fundamentally different operation, Testing, contingency planning, Crew training, maintnenance training, etc. Just training the drivers for the Moorgate branch was a shed load of money.

 

[zwk500]

Well...how many trains are left that aren't ETCS ready? Didn't the decision to start preparing new stock for that happen nearly 20 years ago now?

Believe it was only included in new specifications by default from 2018 in the UK. There's an awful lot of trains built before that currently in servce. And ETCS-ready isn't ETCS-fitted. Given how much the baselines can change, and how technology advances, it's certainly not out of the question that fitting ETCS to a 20-year old "ETCS-ready" train wouldn't be "plug and play".

You've got the vast majority of the freight fleet, for one.

I'd think you'd prioritise fitment of trains over the infrastructure changes, so that the bulk of the track never actually has to be dual fitted.

In an ideal world, although inevitably there will be some areas where you cannot avoid dual fitment. Phasing plans for ECDP and TRU that have been released publicly (some posted on here) have indicated the first schemes in the area will be a transitional

Anything on the infrastructure side seems to be far more expensive than the rolling stock side these days.

Maybe, although that doesn't mean the rolling stock side is peanuts. The rolling stock element of ECDP is hardly cheap as chips.

 

[D365]

Maybe, although that doesn't mean the rolling stock side is peanuts. The rolling stock element of ECDP is hardly cheap as chips.

You can say that again! Not least with the fitment programmes for all 'freight' locomotive classes, aside from Class 43 and Class 66, currently postponed indefinitely.

 

[Annetts key]

ETCS would have determined the braking curve needed to the end of the movement authority (i.e. WH154) and would have warned if it wasn’t complied with, then intervened (brakes) if it was exceeded beyond specific levels. ETCS has to be the way forward.

I agree.

The ATP aspects of ETCS/ERTMS provide the following benefits over just extending TPWS fitment:

• Continuous supervision of permissible line speed including permanent speed restriction and junction speed limits
• Continuous supervision of maximum train speed (maximum speed for the type of rolling stock/traction)
• Continuous supervision of limit of authority (so trains cannot pass signals at red except at a low speed after having stopped) for all main aspect signals, not just those protecting junctions like the current typical TPWS fitment.

Depending on which level is fitted, there are potentially other operational benefits.

== Doublepost prevention - post automatically merged: 24 Jun 2026 ==

Although could be mitigated as @Signal Head suggests by requiring TSS & OSS fitment only at signals normally approached from routes using Approach Control/Release.

However, it is a 'when do you draw the line' question as you could argue similar risks would apply to signals protecting stations, or where the signal is not visible from a platform.

That's the problem with a non-continuous supervision (*) system like TPWS, where it's only fitted to selected signals. Plus, it does not solve the other issues (speeding over points or around curves). And if you do add TPWS to the signals as defined above, what happens when the next train collision occurs somewhere else, where TPWS has not been fitted to a signal?

* - by this I mean the difference between each individual TPWS where they all operate independently and an ATP system (including ETCS/ERTMS) where there is an on-board computer that knows what the signal is showing / the limit of the movement authority / distance, the current train speed and hence can work out the braking curve that should stop the train from passing a signal at red / danger or exceeding the movement authority.

 

[zwk500]

I agree.

The ATP aspects of ETCS/ERTMS provide the following benefits over just extending TPWS fitment:

• Continuous supervision of permissible line speed including permanent speed restriction and junction speed limits
• Continuous supervision of maximum train speed (maximum speed for the type of rolling stock/traction)
• Continuous supervision of limit of authority (so trains cannot pass signals at red except at a low speed after having stopped) for all main aspect signals, not just those protecting junctions like the current typical TPWS fitment.

Depending on which level is fitted, there are potentially other operational benefits.

Yes, ETCS is the best option from an operational viewpoint.

That's the problem with a non-continuous supervision (*) system like TPWS, where it's only fitted to selected signals. Plus, it does not solve the other issues (speeding over points or around curves). And if you do add TPWS to the signals as defined above, what happens when the next train collision occurs somewhere else, where TPWS has not been fitted to a signal?

Extremely valid points, and these are why I'm waiting for further reports from the RAIB to see if there's likely to be any changes coming out of this.
Sadly I think the "proper" fix (more funding to accelerate ETCS rollout) is unlikely to come about until ETCS installation costs are brought down.

 

[Annetts key]

Out of curiosity, will the rollout of ERTMS and ETCS eventually make TPWS redundant or can it simply not fulfil the same role? Even it can is the planned rollout simply too slow to not do anything in the mean time? That's the only question I have about whether it should all signals should be fitted with it.

I'm not sure why would you pay for ETCS L1LS to be installed and then maintain the existing train protection systems?

Surely it would be better to accelerate ETCS fitment than maintain multiple train protection systems?

Once all trains that operate on a line are fitted, AWS and TPWS will become redundant. ATP and ETCS/ERTMS fully replaces the functionally of AWS and TPWS. However, until this time, both AWS and TPWS would have to remain operational.

When no non-ETCS/ERTMS fitted trains remain on a route, all the AWS and TPWS equipment can be removed.

It should be noted that TPWS is not particularly cheap to maintain either. The failure rate on the area where I worked was much worse than the much older GWML ATP. I don't know about ETCS/ERTMS equipment. A "passive" Eurobalise is easy to maintain.

I think I’m right in saying this is the first actual accident where non-fitment of TPWS to certain signals has been a factor? Can anyone point to any others, or any known near-misses?

There will have been incidents that thankfully did not become accidents which ATP or ETCS/ERTMS would have prevented. Some will have been prevented by TPWS (if fitted). But the industry does not publish all this data as far as I am aware.

Certainty over the years that I worked on the railway, I was involved in plenty of internal railway investigations into allegations against the signalling system, some of which did come down to driver error. Thankfully, no one was hurt in any of these.

I presume the list of signals passed at red without permission is still published?

We also don't know how many incidents or accidents the two different ATP systems may have prevented (unless someone can find the data).

 

[HSTEd]

You can say that again! Not least with the fitment programmes for all 'freight' locomotive classes, aside from Class 43 and Class 66, currently postponed indefinitely.

Forgive me, but isn't Class 43 the HST, unless there is an ETCS programme for Warships?

Doesn't the Class 66 fleet represent a rather large portion of the entire freight fleet in of itself?

 

[PhilipRdNGauge]

If this line had been protected by TPWS what sort of speeds would the grids for the the junction have demanded, given that the signal was at yellow. I know this is a high speed mainline, but if you see a yellow you should be expecting a red at the next, so would TPWS slow you right down & keep you there until the next set of grids.

 

[DNCharingX]

Something such as a separate tone for a red signal may have prevented this incident.

 

[Annetts key]

If this line had been protected by TPWS what sort of speeds would the grids for the the junction have demanded, given that the signal was at yellow. I know this is a high speed mainline, but if you see a yellow you should be expecting a red at the next, so would TPWS slow you right down & keep you there until the next set of grids.

The discussion is what would have happened if the signal that was red (WH154) had been fitted.
For just the bare minimum, a TPWS Train Stop System (TSS), the "grids" (officially "loops") are fitted just past the signal. This would apply the train brakes as soon as the train passed the red signal.

If the additional TPWS Over Speed System (OSS) was also fitted, another set of loops is fitted on approach to the signal. If a train is travelling too fast, they trip the brakes. The intention is not to stop a train at line speed from passing the signal, but to slow it enough that any collision is at a much lower speed than if only an TSS is fitted.

 

[swt_passenger]

What is the normal aspect sequence for a train leaving platform 1 and crossing over to the Up Fast running under effectively clear signals?

Given the divergence speed is 50mph for a US-EF move WH356 will have approach control but it only has one divergent route and divergent speed differential is 40mph so i'm not sure that necessitates approach control from Red. So it was more than likely single/double yellow with position 4 junction indicator so subconsciously used to receiving horn warnings in this area. Anyhow RAIB report will disclose all plus what the train warnings being received were.

I’m sure it’s explained in the main thread that the route over the crossover from WH356 definitely is approach release from Red.

 

[YamYamDave]

Sadly, for the railway, this feels like a repeat of the downfall of Railtrack all over again but this time signal tech not rail maintenance. Any politician can now latch onto this and use it to drive something GBR related, but not necessarily for the “greater good”

 

[sharpener]

Something such as a separate tone for a red signal may have prevented this incident.

We know it didn't prevent it because the incident took place.

If you meant "might have prevented the accident" then that is a possibility (I am assuming you are talking about what audio warning was presented to the driver of 1H46) but I thought this thread was about whether extending the usage of AWS/TPWS in their present form would have helped. Developing and rolling out a system with different warning tones for different signal aspects would cost an immense amount of money which would be better spent on rolling out ETCS faster.

AFAIK the current lineside equipment is mostly fed with binary "restrictive aspect/no restrictive aspect" information so the information required to do what you are suggesting would need to come from further back in the signalling infrastructure. S & T ppl will no doubt explain why I am wrong or add more detail.

 

[Harpo]

Sadly I think the "proper" fix (more funding to accelerate ETCS rollout) is unlikely to come about until ETCS installation costs are brought down.

Which is sad. Although the Cambrian pilot came in at just over £100m including train fitment 15 years ago, it was at a time when ETCS was projected to save a third of the typical cost of £250k per-unit, usually per signal in a scheme. Today, as I understand it, that per unit cost for conventional re-signalling is around £400k.

I do wonder how much increased cost has come from the impacts to supplier continuity of renewals ‘pipeline’ micro-management.

 

[Egg Centric]

Poor analogy warning: This to me is like seeing a roulette wheel, thinking "hmm I'd best not bet because I'd lose money but I really like the number 9", the 9 coming up, and then remortgaging your house to put everything on 9 for the next spin. Kinda sorta.

Has anything changed that affects the previous assessments that TPWS everywhere is poor value for money? If not - and it's too early to say presumably - then no way, not for all signals.

What is more likely I assume is that there may be a case for this particular set of circumstances, ie approach release subsequently followed by rapid acceleration, given the increasing performance of modern stock and a few incidents now like this (eg at Peterborough twice and a bit further up ECML) that the risk assessment may be wrong now. But I think this is probably fairly clear to the big boys and you also have to balance that against shorter assumed lifetime with ETCS coming, so I await their conclusion with interest.

 

[Annetts key]

If you meant "might have prevented the accident" then that is a possibility (I am assuming you are talking about what audio warning was presented to the driver of 1H46) but I thought this thread was about whether extending the usage of AWS/TPWS in their present form would have helped. Developing and rolling out a system with different warning tones for different signal aspects would cost an immense amount of money which would be better spent on rolling out ETCS faster.

AFAIK the current lineside equipment is mostly fed with binary "restrictive aspect/no restrictive aspect" information so the information required to do what you are suggesting would need to come from further back in the signalling infrastructure. S & T ppl will no doubt explain why I am wrong or add more detail.

The information available at the location cupboard next to or near to the signal very much depends on the actual installation.

In existing conventional relay based interlocking systems, the following are possible:

The local location cupboard has control relays (or repeat relays) for the aspects. If there are enough spare contacts available, then these spare relay contacts can feed signals to other equipment so that it knows exactly which aspect is lit.

I should mention that for a four aspect signal, the relevant relays are:
DR or DPR
HHR or HHPR
HR or HPR

There are some circuit variations, but typically:

If the HR is de-energised the signal is red,
If the HR is energised and HHR and the DR are de-energised the signal is a single yellow,
If the HR and the HHR are energised and the DR is de-energised the signal is a double yellow,
If the HR, HHR and the DR are all energised, the signal is a green.

If the local location cupboard does not have any or all of the control relays (or repeat relays) for the aspects, the aspects may therefore be fed at 110V AC from either the interlocking if nearby or from another location cupboard.

For TPWS in the later case, new multicore cables have / had to be run in that went to where the control relays were.

The GWML ATP scheme got around this problem by using current transformers - more details in this topic.

For more modern computer based interlocking, if the existing signal module in the local location cupboard does not have enough outputs free (these are binary outputs), then an additional signal module would be needed.

I do agree though, a new system is not practicable given that we already have TPWS and ETCS/ERTMS is supposed to be the future of train protection systems.

 

[Doubleyellow]

Out of curiosity, will the rollout of ERTMS and ETCS eventually make TPWS redundant or can it simply not fulfil the same role? Even it can is the planned rollout simply too slow to not do anything in the mean time? That's the only question I have about whether it should all signals should be fitted with it.

ETCS Level 2 (L2) (which is where we get the most safety benefit) is a "continuous" protection system, the line has to be equipped with it (balises which confirm position and speed through distance travelled) and there needs to be a radio block centre (RBC) for the onboard kit on the trains to "talk" to. The onboard kit "talking" to the lineside (via radio) gets information about upcoming end of movement authorities, or upcoming changes in allowed line speed. All the time it is "asking itself" "how fast am I going?" "what is the gradient here?" and calculating "how long (far) will it take to stop/ slow down if I put the emergency brakes on (+safety margins included in the calculation) here?" (this final one is based on what we tell the onboard about how good this specific train's brakes are). This calculation is sometimes called the "lookahead". There's an offset to the braking lookahead which gives warnings. So for a train approaching an end of movement authority (e.g. associated with an occupied section, with the rear of another train) as you get closer the driver will first get braking indications, then warnings if they don't brake enough; and if the onboard kit calculates that it would have to apply brakes now to stop in time for the end of movement authority - that's what it does.

Some Level 1 implementations are much more like TPWS (associated with a given signal location), and fitment on trains and railways can pave way for future L2 (and beyond) implementation (but as a poster mentions above not much boost in safety benefit)

L2 can give some performance benefits because it will "allow" a train to come closer to the train in front as it gets slower, the "separation" can also be based on the train's actual braking (where eg OSS spacing on a scheme design might have to be based on the "worst" train operating in the area); but you can also "lose" performance because of all the safety margins applied - so sometimes for an experienced driver the ETCS will ask you to start braking before you might normally.

Currently the decision as to whether TPWS (TSS and /or OSS) are fitted at a given signal is based on a risk assessment - as mentioned by posters above, those risk assessments could be revisited, especially for layouts similar to this one given the long timescales in ETCS roll out.

 

[littlerock]

As an interested lay person can I ask a question ? I have seen several reports today which say the Intercity train was stopped because the AWS had triggered its brakes. Isn't AWS an early warning system? Would it be TPWS which would trigger the train's brakes ?