Do the railways have over-speed detection (not just on approach to signals at danger)

[hexagon789]

The 195s and 331s do likewise. They are restricted to 110mph, if you get to that speed you won't be driving again for a while and you won't be allowed any tea or biscuits while you are interviewed afterwards as you will be waiting for medscreen.

I thought they were both 100mph units?

 

[Shaw S Hunter]

I thought they were both 100mph units?

In general terms while a procurement spec will require a certain speed capability the acceptance testing will require a 10% overspeed trial so the technical design spec will inevitably be for a higher max speed than actually required by the customer. An almost inevitable outcome of our strong aversion to risk.

 

[HSTEd]

Does the Channel Tunnel mode still allow speeds up to 200km/h?

No, however since the network code is defined in the packets, as long as some speeds exist in multiple modes, you can use those speeds to translate between modes on the fly.

So the Channel Tunnel mode tops out at 160km/h, and assuming 160km/h is a valid speed in the other mode as well, you can transition seamlessly from one system to the other on the fly.
This is how it works at the ends of the Channel Tunnel.

So you could build a system where we use the Channel Tunnel modes and then use one of the other modes to get the 110/125mph speed codes.

But I think LZB is probably the better system as it avoids this limitation entirely.

Even with SSI there is a cost to blocks, you still need the track circuit or axle counter equipment itself plus the modules that interface to the interlocking. The interlocking itself also has a limit on the number of track circuits, so having more of them may trigger a need for extra interlockings. These limits would also apply to blocks in ETCS, although modern CBIs are less restricted than the original SSI.

Indeed there are, but my understanding is that SSIs are drastically more compact than the traditional relay systems that came before, in that they now fit inside buildings rather than being the size of them.

Adopting a proprietary ATP system in the 1990s would have cost far more than TPWS for little extra safety benefit and for an operational disbenefit as well as being tied to a single supplier. That was the conclusion of the studies into the two ATP systems actually adopted (some of which I contributed to). And we would have been in a technological dead end now with all European countries moving towards ETCS instead. TPWS was of course also a bespoke system, but was at least designed to be easily installed on British traction and signaling rather than requiring complex extra interfaces as the BR ATP systems did.

Some operational disbenefits, but also operational benefits.
For example speed limits would be entirely free of sighting restrictions, which would avoid the banner repeaters sprouting like mushrooms and would substantially improve the economics of regional railway speed increases - which are critical if we are going to make the railway a serious transport system capable of helping us meet our carbon reduction targets.

And whilst it is a technical dead end now.... we would have had the benefits of 20 years of cab signalling.....

But I will yield to your superior knowledge of the state of railway technology in the 90s.
However we are where we are now, and it is critical to get ETCS rolled out rapidly if the railway is to become a bigger slice of the transport pie.

 

[hexagon789]

In general terms while a procurement spec will require a certain speed capability the acceptance testing will require a 10% overspeed trial so the technical design spec will inevitably be for a higher max speed than actually required by the customer. An almost inevitable outcome of our strong aversion to risk.

Hmm, I'm aware of the testing to 10% overspeed, but the post suggested 110 was the normal service maximum whereas I had thought that 195s and 331s were 100mph units. Unless they meant the overspeed kicks in at 110

 

[Llama]

In general terms while a procurement spec will require a certain speed capability the acceptance testing will require a 10% overspeed trial so the technical design spec will inevitably be for a higher max speed than actually required by the customer. An almost inevitable outcome of our strong aversion to risk.

195s and 331s have been over 110mph during testing.

They are indeed 100mph units. Northern only have a safety case to run at 100mph. Test trains are exempt under certain circumstances. The CAF units are now set up to disengage traction power at 110mph. Given a steep enough falling gradient they would continue to increase speed, as far as I know there is no 'penalty brake application' as there is on some other units however an overt 'overspeed' warning would be indicated to the driver on the TCMS screen.

The CAF units would need modifications, derogations from group standards etc to be able to legitimately run above 100mph even if Northern had a safety case for faster trains.

There are no circumstances when they would be permitted to run at over 100mph now. If the driver exceeds 103mph the units automatically notify the control dept., apparently by way of a friendly email. The higher the overspeed the more bleak the consequences for the driver would be, in line with industry guidance for an overspeed incident discovered by any other means.

 

[hexagon789]

They are indeed 100mph units. Northern only have a safety case to run at 100mph.

So yes - 100 in service

 

[Llama]

And 100 when not in service. Special exemptions exist for test trains, the units under test are not always Northern's units by that point too.

 

[edwin_m]

Indeed there are, but my understanding is that SSIs are drastically more compact than the traditional relay systems that came before, in that they now fit inside buildings rather than being the size of them.

A SSI is about the size of a fridge, and two or three of them would replace the relays on the equipment floor of a typical powerbox. For example some of the first schemes were Liverpool Street to Bethnal Green with three interlockings, and one for each end of York station.

Extra track circuits might increase a signalling scheme from three to four SSIs say, which isn't such a big deal on its own. But (as well as the costs of the track circuit hardware) there may be more of the complex data preparation and checking needed if interlocking boundaries fall anywhere other than on a simple plain line.

 

[alxndr]

TPWS was of course also a bespoke system, but was at least designed to be easily installed on British traction and signaling rather than requiring complex extra interfaces as the BR ATP systems did.

What are you referring to as the interfaces here? ATPLITs essentially act as a link in the circuit to the signal lamp feed and it's my understanding that they simply detect the current. If there is current then that aspect must be lit and therefore the encoder knows which aspect the signal is showing. Are they really that much more costly?

 

[edwin_m]

What are you referring to as the interfaces here? ATPLITs essentially act as a link in the circuit to the signal lamp feed and it's my understanding that they simply detect the current. If there is current then that aspect must be lit and therefore the encoder knows which aspect the signal is showing. Are they really that much more costly?

For the ATP interface I was thinking more of the traction - certainly on the GWML interfacing to the HSTs was difficult although I think it was easier for the newer units on Chiltern. TPWS replaces the AWS receiver on the existing mountings and re-uses the AWS brake connection, with a small panel for the extra controls.

At least on pre-existing signals, TPWS breaks the lamp proving circuit if it is faulty. That feature wasn't in the original spec - it was assumed that a loop fault coinciding with a SPAD was unlikely enough not to worry about. I wasn't involved in the programme by the time it was added, but I suspect it might have been due to concerns that a loop fault might go undetected for a long time until a SPAD took place.

 

[alxndr]

For the ATP interface I was thinking more of the traction - certainly on the GWML interfacing to the HSTs was difficult although I think it was easier for the newer units on Chiltern. TPWS replaces the AWS receiver on the existing mountings and re-uses the AWS brake connection, with a small panel for the extra controls.

Ah, I didn't consider the traction side of things with having an S&T background. That makes sense.

At least on pre-existing signals, TPWS breaks the lamp proving circuit if it is faulty. That feature wasn't in the original spec - it was assumed that a loop fault coinciding with a SPAD was unlikely enough not to worry about. I wasn't involved in the programme by the time it was added, but I suspect it might have been due to concerns that a loop fault might go undetected for a long time until a SPAD took place.

Failure indications are included on new signals as well, although there's two slight variations in how it's displayed to the signaller depending on whether it's a retrofit or new installation. On retrofits it's cut into the lamp proving circuitry so it appears the same as an unlit signal, but on new installations there's a separate failure indication. The retrofits used the lamp proving circuitry to avoid having to drill holes and run cables through existing panels which could have caused damage.

ATP on the other hand has no signaller indication as to its state; it relies upon the driver reporting an error code if the beacon is faulty.

 

[Dren Ahmeti]

Moving away from in-cab/signaller indications, I’ve definitely heard of stories of management on SE sitting at a well-known speeding location with a radar/speed gun and flagging up any discrepancies with tea and (no) biscuits!

Sometimes, there are also spot checks at TSR/ESR’s to confirm adherence with the lower speed.

 

[PudseyBearHST]

Moving away from in-cab/signaller indications, I’ve definitely heard of stories of management on SE sitting at a well-known speeding location with a radar/speed gun and flagging up any discrepancies with tea and (no) biscuits!

Sometimes, there are also spot checks at TSR/ESR’s to confirm adherence with the lower speed.

There tends to be a Network Rail bloke at ESRs particularly where there is a big difference between the restriction and normal line speed.
And yes, some driver managers thought it would be fun to stand in such locations but the union put a stop to that.

 

[ComUtoR]

And yes, some driver managers thought it would be fun to stand in such locations but the union put a stop to that.

'Fun' I'm not so sure about. The well know place they would stand is notorious for overspeeds getting tripped. Clocking Drivers at those places reduced incidents dramatically.

There has been a return to gunning. Not sure how long it will last.

 

[mark-h]

Can/does the signaling system produce reports on what speeds the trains are running using the track circuits/axle counters?

 

[MarkyT]

A SSI is about the size of a fridge, and two or three of them would replace the relays on the equipment floor of a typical powerbox. For example some of the first schemes were Liverpool Street to Bethnal Green with three interlockings, and one for each end of York station.

Extra track circuits might increase a signalling scheme from three to four SSIs say, which isn't such a big deal on its own. But (as well as the costs of the track circuit hardware) there may be more of the complex data preparation and checking needed if interlocking boundaries fall anywhere other than on a simple plain line.

Typical system architecture for large SSI schemes has been to house all the interlocking racks at the control centre rather than having remote interlockings situated at each major station or junction, which was typical for relay based technology, although there were some large 'geographical' relay schemes in the Midlands that placed all interlockings for a very wide area at the panel and used vital reed FDM transmission to control the remote trackside equipment. There are thus usually many fridge sized cabinets in typical large panels and IECCs that use SSI technology. Each cabinet of the original equipment could address up to 63 TFMs (trackside functional modules - the distributed input/output of the system) on its duplicated data link, each of which can control and monitor up to two signals or point machines. The latest iterations of the SSI family of products from the major suppliers can pack in a much larger control area for each cabinet, addressing many more TFMs (or alternative i/o devices). Interlocking boundaries are not so critical as they were in the early days with each interlocking now controlling a much larger area. The challenge in the early days was 'timing problems' for messages passing between the original cabinets. 63 addresses was a major limitation for very complex layouts like Paddington making selection of 'ideal' boundary locations impossible, and the data prep team for that project had enormous problems trying to get the system to work satisfactorily. To their great credit, they managed to solve all the problems eventually, although there were significant delays as a result, and there had to be a number of changes to the standard programs and hardware as well a new data prep techniques developed. In retrospect, the technology wan't really mature enough for such a high profile project and a relay interlocking, at least for the final approaches from Old Oak Common to the terminal, might have been more appropriate.

 

[LMS 4F]

Don't know if they still use them, but I have seen portable radar speed guns used on railways. But I think a bigger deterrent will be on-train data recorders - if the record is examined, the driver is likely to be "in the sh*t".

I have seen portable speed gun used at Castleford station, I think it was in 1976. I'm sure things have moved on a bit since then.

 

[alxndr]

There tends to be a Network Rail bloke at ESRs particularly where there is a big difference between the restriction and normal line speed.
And yes, some driver managers thought it would be fun to stand in such locations but the union put a stop to that.

Depending on the reason for the ESR it might necessitate a watchman as well. They'd not be there to catch speeding drivers, but if someone was obviously coming through at linespeed they may well report it!

Can/does the signaling system produce reports on what speeds the trains are running using the track circuits/axle counters?

Not as far as I know, but it would be theoretically possible to calculate an average speed over a section by taking the time from section A going occupied to section B going occupied and dividing it by the distance between the two. However, with audio frequency track circuits there's not a defined point at which one section ends and one begins, which tends to be around 20m so that's already some degree of inaccuracy. There then tends to be a couple of seconds of delay for the information to make it back so that's a bit more delay. With axle counters I'm sure it would potentially be possible to record the time taken between something passing over one wheel sensor and then the next, to get a speed over that head but that's not something that's currently recorded.

I'm really not sure that the expense of coming up with such a system would be worth it given the potential for inaccuracies meaning that it would only be useful in serious cases of overspeeding that would probably be identified by an existing method first.

 

[edwin_m]

Depending on the reason for the ESR it might necessitate a watchman as well. They'd not be there to catch speeding drivers, but if someone was obviously coming through at linespeed they may well report it!

As in this case: https://assets.publishing.service.gov.uk/media/5d4048b440f0b60a9f8021cc/R102019_190801_Sandy.pdf

Network Rail’s standards applicable in these circumstances require the immediate implementation of the speed restriction, hourly inspections of the crossing, and a watchman to be appointed on site to observe the crossing and the passage of trains over it.
...
At around 14:31:46 hrs, shortly after passing over the AWS magnet for signal P268 at the end of Sandy station platform, the driver saw the commencement board ahead and realised the speed restriction applied to his train. He ceased applying power and made a full service brake application, but the speed of the train only reduced to approximately 121 mph (195 km/h) as it went over the set of points.
...
The watchman, who was in a van parked adjacent to the line, immediately telephoned the signaller at Peterborough signal box to report the incident.

 

[Belperpete]

A SSI is about the size of a fridge

Depends on the size of your fridge! An SSI cubicle is about 6ft tall - so about the size of a larder fridge, or fridge-freezer.

 

[Belperpete]

63 addresses was a major limitation for very complex layouts like Paddington making selection of 'ideal' boundary locations impossible, and the data prep team for that project had enormous problems trying to get the system to work satisfactorily. To their great credit, they managed to solve all the problems eventually, although there were significant delays as a result, and there had to be a number of changes to the standard programs and hardware as well a new data prep techniques developed. In retrospect, the technology wan't really mature enough for such a high profile project and a relay interlocking, at least for the final approaches from Old Oak Common to the terminal, might have been more appropriate.

Not sure I would agree with that. Newcastle station was resignalled with SSI before Paddington, with a comparably complex layout, without significant problems. I don't recall that there were significant changes to software and hardware between Newcastle and Paddington. I was one of the Newcastle data team sent to Reading to advise them of "lessons learnt", and the thing we stressed above all was NOT to put in horizontal boundaries. We were therefore horrified to learn later-on that they had not only gone for horizontal boundaries, but for multiple and castellated horizontal boundaries (the most complex kind), and so were not entirely unsurprised when they ran into major problems with the data. Coincidentally, I was part of the project team that removed these horizontal boundaries from the station throat as part of the recent(ish) Paddington resignalling!

To be honest, they were probably forced into having horizontal boundaries. The misconception that horizontal boundaries make the railway more reliable took decades to stamp out.

(A "horizontal" interlocking boundary is where you have two or more pairs of running lines, and you allocate each pair of running lines to a separate interlocking, with an interlocking boundary running between the pairs of lines. The theory is that if one interlocking fails, trains can keep running on the other pair of running lines controlled by the interlocking that is still working. That is only true provided that there are no connections between the pairs of running lines, which is patently not the case in Paddington throat! In such situations, failure of one of the interlockings will disable not only its own pair of lines, but also all adjacent line(s) there is a connection to! So horizontal boundaries actually make the railway LESS reliable, not more. They also significantly increase the data complexity and volume.)

 

[Belperpete]

it would be theoretically possible to calculate an average speed over a section by taking the time from section A going occupied to section B going occupied and dividing it by the distance between the two.

Something similar is sometimes provided with level crossings, where a train is timed between a pair of treadles. If the train passes too quickly, then it is likely that it will overrun the signal protecting the level-crossing, and so the crossing is triggered. On all the schemes that I know of, this has had to be done in circuits, as even "next generation SSIs" are deemed too slow.

With SSI, there are a series of steps that a change of track-circuit state has to go through, from being input by the trackside input processor, to passing that information to the central processor, the interlocking then processing that information, and so on. If there were a consistent delay, then that could be taken into account in the calculation, but unfortunately it is a bit like driving through a series of traffic lights: if you are lucky, they are all green and you go straight through with no delay, but if you are unlucky then you get held-up at every stage. Likewise with SSI, the information may get straight through, but in practice at each step there will be a variable delay something between zero and the maximum.

Unfortunately SSI is 1970s technology, and so the system runs at 1970s speeds. The worst-case delay can be several seconds. It would have to be a VERY long section, or a very low speed, for a 0-2 second delay on both enter and exit to be insignificant.

Even "next generation" SSIs currently still use traditional SSI trackside equipment, and so are limited to their speeds. Things will improve with the introduction of "next generation" trackside modules, but even so I am not sure the system timings will be good enough to allow for accurate calculation of train speeds.

 

[alxndr]

On all the schemes that I know of, this has had to be done in circuits, as even "next generation SSIs" are deemed too slow.

With SSI, there are a series of steps that a change of track-circuit state has to go through, from being input by the trackside input processor, to passing that information to the central processor, the interlocking then processing that information, and so on.

As I said, it could be done, but would be very impracticable. Doing it at circuitry level would increase accuracy but also be more costly. Perhaps if there was a known problem area it would be worth it, but equally you could use an existing technology like TPWS to prevent the overspeed in the first place (as for some PSRs) rather than just reporting it if it does happen. Prevention is always far better than punishment after the fact.

SSI can be tediously slow indeed. Tripped us up a fair few times when we first started working with SSI and testing to the signaller. One of the many reasons I'm a dinosaur and would rather have a proper relay interlocking any day.

 

[Llama]

There was an ESR (possibly daggered too) on the WCML somewhere south of Preston a good few years ago, 20 on the up slow line due to renewals on the fasts. ALO working if I'm not mistaken. I'm pretty sure we were advised at the time that there was speed monitoring in place, I can't remember the wording of the notice but I'm sure I was led to believe it was radar monitoring?

 

[MarkyT]

Not sure I would agree with that. Newcastle station was resignalled with SSI before Paddington, with a comparably complex layout, without significant problems. I don't recall that there were significant changes to software and hardware between Newcastle and Paddington. I was one of the Newcastle data team sent to Reading to advise them of "lessons learnt", and the thing we stressed above all was NOT to put in horizontal boundaries. We were therefore horrified to learn later-on that they had not only gone for horizontal boundaries, but for multiple and castellated horizontal boundaries (the most complex kind), and so were not entirely unsurprised when they ran into major problems with the data. Coincidentally, I was part of the project team that removed these horizontal boundaries from the station throat as part of the recent(ish) Paddington resignalling!

To be honest, they were probably forced into having horizontal boundaries. The misconception that horizontal boundaries make the railway more reliable took decades to stamp out.

(A "horizontal" interlocking boundary is where you have two or more pairs of running lines, and you allocate each pair of running lines to a separate interlocking, with an interlocking boundary running between the pairs of lines. The theory is that if one interlocking fails, trains can keep running on the other pair of running lines controlled by the interlocking that is still working. That is only true provided that there are no connections between the pairs of running lines, which is patently not the case in Paddington throat! In such situations, failure of one of the interlockings will disable not only its own pair of lines, but also all adjacent line(s) there is a connection to! So horizontal boundaries actually make the railway LESS reliable, not more. They also significantly increase the data complexity and volume.)

Thanks for the explanation. So it seems the data preppers were set an unnecessarily difficult task at Paddington. Were the Newcastle boundaries better designed? I remember the obsession with horizontal boundaries in the early days. It seemed to me the wrong emphasis, as a complete interlocking failure or intentional shutdown should be a very rare event indeed compared to partial failures in the trackside systems. Although some routes on an unaffected parallel pair might still be setable without the cross boundary flank conditions etc, the concept is clearly not worth the vastly increased complexity. Pleased to hear it has been finally sorted out in the latest layout changes.

 

[westcoaster]

On more modern units , managers can sit in an office. Type in a unit number and watch in real time what the driver is doing, pressing or moving. Very hi-tech stuff.

 

[pompeyfan]

On more modern units , managers can sit in an office. Type in a unit number and watch in real time what the driver is doing, pressing or moving. Very hi-tech stuff.

correct, they can also download and watch FFCCTV remotely

 

[edwin_m]

Thanks for the explanation. So it seems the data preppers were set an unnecessarily difficult task at Paddington. Were the Newcastle boundaries better designed? I remember the obsession with horizontal boundaries in the early days. It seemed to me the wrong emphasis, as a complete interlocking failure or intentional shutdown should be a very rare event indeed compared to partial failures in the trackside systems. Although some routes on an unaffected parallel pair might still be setable without the cross boundary flank conditions etc, the concept is clearly not worth the vastly increased complexity. Pleased to hear it has been finally sorted out in the latest layout changes.

One end of Newcastle is a lot simpler than Paddington on just about any metric (approach tracks, platforms, points etc). I don't remember if Newcastle had separate interlockings for the two ends like York but I think it would - splitting down the middle of the platforms isn't as easy as plain line but certainly easier than trying to sub-divide the throat.

I think they also split detection of crossovers at Paddington in an attempt to keep one end working if detection on the other end failed.

 

[Belperpete]

Thanks for the explanation. So it seems the data preppers were set an unnecessarily difficult task at Paddington. Were the Newcastle boundaries better designed?

I suspect more by chance than design. It was only after experiences at Newcastle and Paddington that people woke up to the difficulties that horizontal boundaries could cause data production. On early SSI schemes, interlocking boundaries were decided purely on a count of the TFMs. On later schemes, a lot more attention was paid to keeping the interlocking boundaries as simple as possible, and in particular not drawing them through the middle of swinging overlaps!

The Paddington boundaries were particularly complex because they were castellated, which meant routes often passed from one interlocking to another, and then back into the first interlocking. And in some cases, then back to the second interlocking again! And perhaps to yet another interlocking at the exit/overlap. Overlaps could also swing over the castellated boundaries, with similar complexities. This all added up to a considerable volume of cross-boundary data within each interlocking, requiring use of space-saving techniques. And also a considerable amount of information being transmitted across the boundaries, such that they ran short of cross-boundary telegrams, necessitating some novel compression techniques.

Unfortunately another bug-bear that SSI suffered for a long time was the "its only data" syndrome: data preparers were expected to cope with anything that was thrown at them no matter how much or how complex the data required, whereas something that would have required an extra rack of relays in a relay interlocking would have been questioned.

 

[Belperpete]

I don't remember if Newcastle had separate interlockings for the two ends like York but I think it would - splitting down the middle of the platforms isn't as easy as plain line but certainly easier than trying to sub-divide the throat.

I am fairly certain that Newcastle had both horizontal boundaries and vertical boundaries (including through the middle of the station).

I think they also split detection of crossovers at Paddington in an attempt to keep one end working if detection on the other end failed.

Paddington certainly had some separately-detected crossovers, but only some, and only where the crossovers concerned were clear of overlaps. More were added at the recent resignalling. The data for separate-end detection is quite simple.