TPWS Every Signal

[Hitacky IET]

TPWS is only installed at 'high risk' signals and locations. Why has it not been installed at *every* signal, similar to each signal having AWS or every LUL signal having the tripcock system?

No matter where it is, a red signal is rather 'high risk' to me - regardless of the location!

 

[DorkingMain]

One word: cost.

 

[bengley]

On the contrary, on the routes I sign, there are a couple of high risk controlled signals where TPWS is not fitted where you'd think it should be.

Cove LC on the WCML is one of them. No overspeed or train stop loops on a 125mph line, approaching a CCTV level crossing. Quite extraordinary if you ask me!

 

[Dstock7080]

TPWS is only installed at 'high risk' signals and locations. Why has it not been installed at *every* signal, similar to each signal having AWS or every LUL signal having the tripcock system?

There are no trainstops on the Central, Jubilee, Northern, Victoria Lines and parts of Circle District and Metropolitan Lines.

 

[Hitacky IET]

"There are no trainstops on the Central, Jubilee, Northern, Victoria Lines and parts of Circle District and Metropolitan Lines."

Yes, but each one of those resignalled lines has the alternative of a cocktrip, including ATP (Central) or CBTC (Jub/Northern/SSR) - all will stop a train passing a signal at danger.

 

[DorkingMain]

To give the long answer:

The London Underground network is a relatively small network with a huge number of trains passing over it. There's plenty of signals in the UK that get one train a week rolling past them, or are strictly used by freight, or for diversions, etc. I totally agree TPWS should be at every high risk location but it wouldn't be particularly economical to install it in places which have few trains or present little risk of a SPAD occurs. SPADs on a line where the blocks are miles long are very different to a SPAD where a train crosses onto a busy junction.

 

[edwin_m]

I was part of the project that developed the idea of TPWS in the mid-1990s. The ATP systems installed on GW and Chiltern were costly, restrictive of capacity and later became obsolete, and there was a need to find something easier to fit as well as cheaper. All accidents back to about 1968 were analysed to estimate what the casualties might have been if a TPWS-like system had been fitted, along with various other options including the Driver Reminder Appliance. It was concluded that TPWS (fitted only to high risk signals) would save some 70% of the casualties that ATP would have, and experience since then has borne out that prediction.

 

[ge-gn]

On the contrary, on the routes I sign, there are a couple of high risk controlled signals where TPWS is not fitted where you'd think it should be.

Cove LC on the WCML is one of them. No overspeed or train stop loops on a 125mph line, approaching a CCTV level crossing. Quite extraordinary if you ask me!

Not familiar with that area, but it is normal practice that if the protecting signal of a signaller worked level crossing is SPAD the overlap track circuit will initiate a sequence and immediately bring up reds to road users. Perhaps this is the reason (along with cost as mentioned) that TPWS wasn’t installed at such locations.

 

[43066]

TPWS is only installed at 'high risk' signals and locations. Why has it not been installed at *every* signal, similar to each signal having AWS or every LUL signal having the tripcock system?

No matter where it is, a red signal is rather 'high risk' to me - regardless of the location!

Cost, and the principle of diminishing returns. How many lives have been lost on the railway network over the last two decades that installing TPWS at every single signal would have saved?

None!

There is your answer.

 

[bengley]

Not familiar with that area, but it is normal practice that if the protecting signal of a signaller worked level crossing is SPAD the overlap track circuit will initiate a sequence and immediately bring up reds to road users. Perhaps this is the reason (along with cost as mentioned) that TPWS wasn’t installed at such locations.

I suspected that may be the case. It just seems weird as a driver!

 

[Skoodle]

All signals on East London Line are fitted with TPWS. Most also have OSS grids on approach. (Highbury to New Cross, New Cross Gate and Old Kent Road Junction).

 

[kevin_roche]

AIUI the rules also say that train protection can't be made worse than it is already which is why the TPWS had to be upgraded on the GWML for the Crossrail trains. Originally was supposed to be using ETCS L2 (as an overlay) but when that did not happen TPWS had to be upgraded. The crossrail trains don't have the GW/ATP system installed that the other trains on the GWML have.

 

[edwin_m]

AIUI the rules also say that train protection can't be made worse than it is already which is why the TPWS had to be upgraded on the GWML for the Crossrail trains. Originally was supposed to be using ETCS L2 (as an overlay) but when that did not happen TPWS had to be upgraded. The crossrail trains don't have the GW/ATP system installed that the other trains on the GWML have.

The 165/166s don't have GW-ATP. Do the 387s? I suspect not, as they couldn't run to Heathrow until that route got ETCS.

 

[TSG]

TPWS is only installed at 'high risk' signals and locations. Why has it not been installed at *every* signal, similar to each signal having AWS or every LUL signal having the tripcock system?

No matter where it is, a red signal is rather 'high risk' to me - regardless of the location!

Although cost is an issue It isn't as simple as that. There is a limit to how much signalling design, installation and testing resource (not to mention track access) you can call on, even if the finance wasn't limited. It would be irresponsible to use those resources to fit TPWS to signals that show a red signal to a driver once in a decade rather than to reduce risk at level crossings or renew life expired signalling, since they present much greater risks. Bear in mind that those renewals will have at least passive provision for ETCS which is the way to eventually achieve a greater level of safety than what you propose with blanket TPWS

 

[LNW-GW Joint]

On the contrary, on the routes I sign, there are a couple of high risk controlled signals where TPWS is not fitted where you'd think it should be.
Cove LC on the WCML is one of them. No overspeed or train stop loops on a 125mph line, approaching a CCTV level crossing. Quite extraordinary if you ask me!

Not to mention the word "Quintinshill" just to the south.
It will all be resignalled, probably with ETCS, if the present HS2 plans go ahead.

 

[bengley]

I drove over cove today in both directions and particularly on the down line, by the time you've SPADded the protecting signal and gone over the insulated block joint which would indicate you've passed the signal at danger, there would be about one second, probably less, of red road lights showing to road users. Scary.

 

[Domh245]

I drove over cove today in both directions and particularly on the down line, by the time you've SPADded the protecting signal and gone over the insulated block joint which would indicate you've passed the signal at danger, there would be about one second, probably less, of red road lights showing to road users. Scary.

Is that at (close to) line speed, because if so, the 'more likely' spad will involve a slower speed and longer road user red light time as a result

 

[edwin_m]

Can anyone cite any accident where a SPAD at the protecting signal has led to a train hitting road vehicles on a level crossing? I can only think of Moreton-on-Lugg, which was different because the signaler replaced the signal and opened the barriers before the train had reached them, and had the interlocking been to modern standards it would have enforced a timeout before the barriers could be raised.

Having said that traditional signalling practice has been not to treat a level crossing as an obstruction to a train. Until relatively recently the standards allowed an unrestricted approach to a signal just a few yards away. I'm not familiar with the criteria for applying TPWS but it's probable that level crossings aren't taken into account when assessing the risk.

 

[Horizon22]

As others have mentioned cost. I think it was the 1990s when BR did a cost-benefit analysis to protect the most at-risk junctions. And to be fair there hasn't been a serious SPAD resulting in deaths since. Cost-benefit analysis is used in all aspects of life, in safety positions. There's even an an objective cost for human life.

You could theoretically take that argument everywhere - why don't we have hundreds of drivers sitting spare etc.?

 

[edwin_m]

Strictly speaking Moreton-on-Lugg was a SPAD causing a fatality, but from a totally different cause and not prevented by TPWS being fitted.

 

[OVS Flatbottom]

Can anyone cite any accident where a SPAD at the protecting signal has led to a train hitting road vehicles on a level crossing? I can only think of Moreton-on-Lugg, which was different because the signaler replaced the signal and opened the barriers before the train had reached them, and had the interlocking been to modern standards it would have enforced a timeout before the barriers could be raised.

Having said that traditional signalling practice has been not to treat a level crossing as an obstruction to a train. Until relatively recently the standards allowed an unrestricted approach to a signal just a few yards away. I'm not familiar with the criteria for applying TPWS but it's probable that level crossings aren't taken into account when assessing the risk.

Milford, 1978? https://www.railwaysarchive.co.uk/documents/MoT_Milford1978.pdf

There was also an accident at Slaght in Northern Ireland in 1990, but it wasn't exactly a SPAD; the lights failed at an Automatic Open Crossing Locally Monitored, and the Driver's White Light approaching the crossing was correctly extinguished, but the driver of the train - who I think was distracted by chatting to the guard - failed to stop and hit a car on the crossing, killing a passenger on the train and two people in the car. There's a report, but it isn't online as far as I know. (I think in that case the train driver was also tried for manslaughter and acquitted.)

 

[43096]

Cost, and the principle of diminishing returns. How many lives have been lost on the railway network over the last two decades that installing TPWS at every single signal would have saved?

None!

There is your answer.

A follow-up question: when was the last fatal accident that would only have been preventable by ATP i.e. TPWS and/or AWS would not have prevented it? I'm struggling to think of one.

 

[edwin_m]

Milford, 1978? https://www.railwaysarchive.co.uk/documents/MoT_Milford1978.pdf

Yes, that would count as one. The report notes it's the first such accident at an AHB, so unless anyone can come up with any others I think my contention stands that a SPAD causing a collision at a crossing is a rare event. A few years after that accident I was a reasonably regular visitor to Liss, where the level crossing was probably upgraded at the same time in a similar way, and my hazy recollection is that the signal was right at the top of the platform ramp with the crossing just beyond. I've just checked an image search and there is no platform ramp, which may be a later modification as ramps are not now preferred, but the signal is about where the top of the ramp would be with a rather useless section of platform beyond it.

 

[Chris Hamilton]

On the contrary, on the routes I sign, there are a couple of high risk controlled signals where TPWS is not fitted where you'd think it should be.

Cove LC on the WCML is one of them. No overspeed or train stop loops on a 125mph line, approaching a CCTV level crossing. Quite extraordinary if you ask me!

Not an unusual occurrence on the top end of the WCML.

As, neither did Logan’s Road (in the Up direction) up until its recent resignalling. The down only had it by dint of its protecting signal also protecting a junction, and one with a severely reduced overlap at that.

Cleghorn also only has TPWS because all 4 of its protecting signals also protect Lanark Junction.

Also, slightly further north, but still in the former Motherwell SC area, Heatherbell LC doesn’t have any TPWS at all, and I’m fairly certain that it was resignalled to SSI after the initial nationwide introduction of TPWS.

What needs to be remembered is that TPWS only reduces the risk and lowers the potential collision speed, it doesn’t eliminate it in the way that ATP can. If a driver chooses to disregard the signal sequence and pile towards a red signal at 125mph (which would of course be fitted with TPWS+ for such a high speed) then he or she will still be doing a pretty horrific speed as they pass the protecting signal and then the obstruction. I’ve investigated enough SPADs where the overlap was used up and then some, sometimes due to equipment failure (signal returning to danger in the drivers face), sometimes due to operator error (brain fart moment of applying power towards a red aspect, usually after a station stop or being complacent that the signal will change as it always does).

For automatic signals it is of little more use than AWS (excepting the ability to blindly acknowledge the AWS and carry on). The probability of a preceding stopped train being just beyond the Red auto signal (inside the emergency braking distance) is slim. A full AWS initiated brake application will stop a train no different to a TPWS application (give or take a few seconds due to the potential for OSS+ and OSS initiating the application further out). Everything costs money, and someone somewhere does the calculation of risk versus cost. Installing TPWS to every signal would require a team of designers working flat out to alter the drawings of every signal, teams of installers to install the equipment, not to mention the added maintenance cost and potential for introducing failures. The incidence of rear end collisions from mistakenly passed Red auto signals is very very low, and remains so.

TPWS does have other uses also, and it is very successful in enforcing permanent speed restrictions (where only the OSS is installed)

 

[MarkyT]

I was part of the project that developed the idea of TPWS in the mid-1990s. The ATP systems installed on GW and Chiltern were costly, restrictive of capacity and later became obsolete, and there was a need to find something easier to fit as well as cheaper. All accidents back to about 1968 were analysed to estimate what the casualties might have been if a TPWS-like system had been fitted, along with various other options including the Driver Reminder Appliance. It was concluded that TPWS (fitted only to high risk signals) would save some 70% of the casualties that ATP would have, and experience since then has borne out that prediction.

Another critical resource was time and there was the perception that UK was falling seriously behind many other European railways. Some were streets ahead like Sweden with full ATP everywhere already since the 80s, but by the late 90s, most major railways had basic limited supervision systems capable of some train stop and speed supervision functionality as well as distant warning (which we already had with AWS of course). In Europe these were often applied on a risk assessed basis, but with varying criteria. The simplicity of TPWS meant within two or three years UK was able to implement the most safety impact within the shortest time and largely catch up with our neighbours. The tech is a bit old and basic and not intrinsically failsafe, but it has proved very reliable and most importantly has definitely saved many lives and injuries from potential junction collisions.

 

[JN114]

Isn't every brake intervention caused by the TPWS intervention a possible major incident averted ?

Without context, possibly. In practice a lot of activations the driver is already braking for the signal, so there’s no saying whether they’d even have passed the signal, never mind by enough to cause a major incident.

Take for example on GW, where the IETs are fitted with TPWS and ATP.

ATP is continually computing the “safe” speed for the train to be doing approaching the signal, with distance to go etc; and it’ll warn the driver to brake harder if they’re cutting it fine, and it will step in and apply the brakes itself if the driver doesn’t improve their retardation.

But despite that, we get IETs tripped up on TPWS approaching signals. ATP hasn’t warbled because on the continually computed curve the train is fine, but TPWS wants the driver to be doing less - and the set speeds are, officially at least, blind. They’re not divulged to driving grades, at least at my TOC as “it’d be like offering a sugar lump to a racehorse”.

 

[Bald Rick]

A follow-up question: when was the last fatal accident that would only have been preventable by ATP i.e. TPWS and/or AWS would not have prevented it? I'm struggling to think of one.

Strictly speaking, Southall ? AWS fitted, but didn’t prevent it (and we’ll never know if it would have done, had it been operable). TPWS wouldn’t have prevented it either, but probably would have reduced the collision speed.

 

[43096]

Strictly speaking, Southall ? AWS fitted, but didn’t prevent it (and we’ll never know if it would have done, had it been operable). TPWS wouldn’t have prevented it either, but probably would have reduced the collision speed.

The AWS was defective, so it was an AWS preventable accident. It was a pre-1910 (on the Great Western at least) accident in 1997.

 

[43066]

Strictly speaking, Southall ? AWS fitted, but didn’t prevent it (and we’ll never know if it would have done, had it been operable). TPWS wouldn’t have prevented it either, but probably would have reduced the collision speed.

Yep I’d thought that, until reading that the AWS was also isolated. Of course that would be absolutely unthinkable these days, but was apparently permissible practice at the time!

 

[Bald Rick]

The AWS was defective, so it was an AWS preventable accident. It was a pre-1910 (on the Great Western at least) accident in 1997.

Yes I know, but we don’t know whether having AWS operable would have prevented it. The driver missed 2 cautionary signals - we will never know if he would have cancelled the AWS anyway.

Anyway, back on topic, AWS didn’t prevent either Purley or Wembley, and I doubt very much if TPWS would have done either, although, again, it would have probably reduced the collision speed in both cases.