Could alterations be made to increase stopping distance at critical locations?

[L&Y Robert]

Moderator note: Split from

Collision and derailment near Salisbury (Fisherton Tunnel) 31/10/21

Mod note: Please take care when posting; any speculation should be reasonable, proportional and clearly identified as such. As this incident is being reported in the mainstream media, we need to ensure speculation posted here isn't picked up and reported as fact. Additionally, this is a...

www.railforums.co.uk

If the Salisbury collision was wheel-slide (not entirely certain yet) could this sort of accident be avoided by simply re-configuring the system to allow more stopping distance in critical locations? - and to be triggered automatically when conflicting movements are likely?

I thought that Lax 54 up-thread (post 864) had the obvious answer. Keep 'em well apart, not just one red apart.
And what about catch-points and sand drags?

 

[Horizon22]

Keep 'em well apart, not just one red apart.

Surely that's the reason that there are yellow and in some places double yellow signals?

 

[quattromatt]

If it was wheel-slide (not entirely certain yet) could this sort of accident be avoided by simply re-configuring the system to allow more stopping distance in critical locations? - and to be triggered automatically when conflicting movements are likely?
I thought that Lax 54 up-thread (post 864) had the obvious answer. Keep 'em well apart, not just one red apart.
And what about catch-points and sand drags?

If it is wheelslide then the lesson is manage the foliage better.

 

[Signal Head]

If it was wheel-slide (not entirely certain yet) could this sort of accident be avoided by simply re-configuring the system to allow more stopping distance in critical locations? - and to be triggered automatically when conflicting movements are likely?
I thought that Lax 54 up-thread (post 864) had the obvious answer. Keep 'em well apart, not just one red apart.
And what about catch-points and sand drags?

It can, and is in some places. It can be done by requiring a signal to have its route set before a route can be set up to it, the effect of which is to hold the train one section further back. This may be relaxed by an 'approach release' condition to (attempt to) check that the train is under control at the rear signal, which is then allowed to clear to a yellow aspect.

That may, or may not have helped in this instance, it probably depends on where the slide started - if the SW train passed 29 signal fully under control at an appropriate speed, and then lost adhesion during the final brake application to stop at 31, then I can't see that putting such controls on 29 would do much good.

 

[High Dyke]

It can, and is in some places. It can be done by requiring a signal to have its route set before a route can be set up to it, the effect of which is to hold the train one section further back. This may be relaxed by an 'approach release' condition to (attempt to) check that the train is under control at the rear signal, which is then allowed to clear to a yellow aspect.

That may, or may not have helped in this instance, it probably depends on where the slide started - if the SW train passed 29 signal fully under control at an appropriate speed, and then lost adhesion during the final brake application to stop at 31, then I can't see that putting such controls on 29 would do much good.

I am currently working a location not dissimilar to the Laverstock Jn. / Salisbury layout: i.e triangular junction (though the location of a tunnel has no causal effect). The two converging routes at the junction are protected by approach controlled signals. So a train from one route may be signalled without restriction, but a train on the converging route would be allowed to approach the junction by the signals clearing on approach control, thus reducing the speed of approach.

 

[MarkyT]

I am currently working a location not dissimilar to the Laverstock Jn. / Salisbury layout: i.e triangular junction (though the location of a tunnel has no causal effect). The two converging routes at the junction are protected by approach controlled signals. So a train from one route may be signalled without restriction, but a train on the converging route would be allowed to approach the junction by the signals clearing on approach control, thus reducing the speed of approach.

Probably a 'restricted' overlap clear of the junction in that case, requiring a 'warning' class route with a delayed yellow aspect when the junction is blocked. At Salisbury, SY31's overlap, 'reduced' at 172yds (from diagram upthread), is usually considered acceptable for the 50mph approach speed (subject to risk assessment) so an unrestricted 'main' class route approach can apply.

 

[Taunton]

The "that overlap's OK, it's only 50mph there" doesn't seem to allow for the line speed to shortly beforehand having been 90mph, and trains have to be braking continuously from the outer distant at YY to even hit the start of the 50mph limit correctly. Interesting to see what the RAIB will make of this - they criticised a similar combination at a Watford accident some years ago.

 

[Watershed]

The "that overlap's OK, it's only 50mph there" doesn't seem to allow for the line speed to shortly beforehand having been 90mph, and trains have to be braking continuously from the outer distant at YY to even hit the start of the 50mph limit correctly. Interesting to see what the RAIB will make of this - they criticised a similar combination at a Watford accident some years ago.

But even if the overlap had been standard, it would have made little difference to the outcome. We are talking about a difference of a few yards here; realistically, something like double red protection would have been needed to prevent the accident, and that is simply unacceptable in terms of headway/capacity.

 

[MarkyT]

The "that overlap's OK, it's only 50mph there" doesn't seem to allow for the line speed to shortly beforehand having been 90mph, and trains have to be braking continuously from the outer distant at YY to even hit the start of the 50mph limit correctly. Interesting to see what the RAIB will make of this - they criticised a similar combination at a Watford accident some years ago.

Not sure if it's exactly the same as current design principles, but when I was designing signalling layouts some decades ago, the speed at 440 yds on approach to the signal concerned was able to be used to reduce the overlap, if expedient. Remember it is a 40-year old installation at Salisbury. On a new build layout today, the risk assessment requirement may have triggered further mitigation. Note from the TPWS tabulation above that the conflict point from SY 31 is stated as 155m and an overspeed intervention on approach (just after Laverstock N) would bring a train to a stand within this distance, given normal braking performance. Furthermore, the 90>50 PSR commencement, as a reduction of more than one third, is protected by its own permanently activated overspeed installation, just before the previous signal, SY29. The issue here appears to be braking performance in specific conditions.

== Doublepost prevention - post automatically merged: 7 Nov 2021 ==

But even if the overlap had been standard, it would have made little difference to the outcome. We are talking about a difference of a few yards here; realistically, something like double red protection would have been needed to prevent the accident, and that is simply unacceptable in terms of headway/capacity.

I expect SY31 is positioned where it is to ensure the longest typical train formation waiting at Tunnel Jn doesn't block back over Laverstock N Jn, preventing another movement over the loop. It is about 220m standage, measured on Google Earth. If SY31 was moved back to provide a standard full overlap clear of Tunnel Jn then a 9-car 159 would probably not fit. In further complexity, the signal's overspeed TPWS installation and possibly its AWS ramp might have to move east of Laverstock N Jn in that case, which would necessitate measures to suppress warnings and interventions for moves to and from Laverstock Loop.

 

[DelW]

Not sure if it's exactly the same as current design principles, but when I was designing signalling layouts some decades ago, the speed at 440 yds on approach to the signal concerned was able to be used to reduce the overlap, if expedient. Remember it is a 40-year old installation at Salisbury. On a new build layout today, the risk assessment requirement may have triggered further mitigation. Note from the TPWS tabulation above that the conflict point from SY 31 is stated as 155m and an overspeed intervention on approach (just after Laverstock N) would bring a train to a stand within this distance, given normal braking performance. Furthermore, the 90>50 PSR commencement, as a reduction of more than one third, is protected by its own permanently activated overspeed installation, just before the previous signal, SY29. The issue here appears to be braking performance in specific conditions.

== Doublepost prevention - post automatically merged: 7 Nov 2021 ==


I expect SY31 is positioned where it is to ensure the longest typical train formation waiting at Tunnel Jn doesn't block back over Laverstock N Jn, preventing another movement over the loop. It is about 220m standage, measured on Google Earth. If SY31 was moved back to provide a standard full overlap clear of Tunnel Jn then a 9-car 159 would probably not fit. In further complexity, the signal's overspeed TPWS installation and possibly its AWS ramp might have to move east of Laverstock N Jn in that case, which would necessitate measures to suppress warnings and interventions for moves to and from Laverstock Loop.

Would it ever be permissible and/or practical to install catch points at a location like this? Just thinking that diverting the 159 into the cutting side, while not eliminating a derailment, might have kept it clear of the GWR train. That could have restricted the damage to just the 159, and maybe reduced the number of passenger injuries, and halved the number of coaches likely to be scrapped.

 

[Watershed]

Would it ever be permissible and/or practical to install catch points at a location like this? Just thinking that diverting the 159 into the cutting side, while not eliminating a derailment, might have kept it clear of the GWR train. That could have restricted the damage to just the 159, and maybe reduced the number of passenger injuries, and halved the number of coaches likely to be scrapped.

Catch points are difficult to sensibly introduce to a converging triangular junction like this.

If you used them to direct the train towards the Up line, you'd have to prevent any Up movements towards Basingstoke. Unless you're suggesting that the catch points would default to leading towards the Up line, but could be swung back if there is an Up train to Basingstoke? In which case your protection only exists if there isn't an opposing train approaching.

 

[DelW]

Catch points are difficult to sensibly introduce to a converging triangular junction like this.

If you used them to direct the train towards the Up line, you'd have to prevent any Up movements towards Basingstoke. Unless you're suggesting that the catch points would default to leading towards the Up line, but could be swung back if there is an Up train to Basingstoke? In which case your protection only exists if there isn't an opposing train approaching.

I was thinking of them deflecting the train towards the down side cess, where it would hopefully come to rest against the cutting side. That does mean they'd have to be installed beyond signal 31 (obviously), but before reaching the cutting for the Romsey lines. I don't know if there's room to do that at this location, but I was also interested in the general case as to whether they'd be an option or not.

 

[MarkyT]

Would it ever be permissible and/or practical to install catch points at a location like this? Just thinking that diverting the 159 into the cutting side, while not eliminating a derailment, might have kept it clear of the GWR train. That could have restricted the damage to just the 159, and maybe reduced the number of passenger injuries, and halved the number of coaches likely to be scrapped.

A simple derailing trap wouldn't be acceptable but a short 'overlap spur' leading to a friction bufferstop or in the old days a sand drag might be an option. Normally these are only provided where there is no room for any kind of sensible clear overlap, yet flexibility demands the ability to run towards the signal while the junction ahead is blocked. I would suggest treating the existing overlap as 'restricted', with an additional 'warning' class route from SY29 to SY31, enforcing a delayed yellow clearance at SY29. The unrestricted 'main' route would also still be available but would have an extended overlap through the junction. Whether the warner could have reduced speed further back sufficient to prevent the slide past, or limit its extent enough to avoid the collision, I really couldn't say. I want to reiterate this is a 40-year-old layout in this form. If it was inherently dangerous, surely that would have been uncovered before. Have there been any worrying SPADs at this junction in the past, or any near misses?

 

[thirdraildrive]

In the area I drive in, mainly 4 aspect, there seems to always be a system in place whereby a conflicting movement is double protected by two reds. So you'll get 2Y, 1Y, R. The red will clear to a 1Y on approach thus having slowed you right down, then you finaly stop at the red actually protecting the movement. Question... is this a national thing or location specific? As annoying as it can be, it does a very good job of reducing the speed on approach to such conflicts and I would have thought especially at a location like this this would be the method of signalling?

 

[bengley]

In the area I drive in, mainly 4 aspect, there seems to always be a system in place whereby a conflicting movement is double protected by two reds. So you'll get 2Y, 1Y, R. The red will clear to a 1Y on approach thus having slowed you right down, then you finaly stop at the red actually protecting the movement. Question... is this a national thing or location specific? As annoying as it can be, it does a very good job of reducing the speed on approach to such conflicts and I would have thought especially at a location like this this would be the method of signalling?

That's a very standard thing nationally

 

[Tomnick]

In the area I drive in, mainly 4 aspect, there seems to always be a system in place whereby a conflicting movement is double protected by two reds. So you'll get 2Y, 1Y, R. The red will clear to a 1Y on approach thus having slowed you right down, then you finaly stop at the red actually protecting the movement. Question... is this a national thing or location specific? As annoying as it can be, it does a very good job of reducing the speed on approach to such conflicts and I would have thought especially at a location like this this would be the method of signalling?

It's a control applied to mitigate against a specific risk or permit something like a short overlap, so it's location-specific but not unique or particularly unusual. My favourite example on my routes, I understand that they're there because TPWS alone might not stop a train from linespeed short of the fouling point if it passes the protecting signal at danger - effectively this arrangement allows the train to be proved under control at the signal in rear (which won't step up to a yellow until the train's passed over the overspeed loops associated with that signal) and thus slowed down sufficiently to ensure that TPWS can do its job properly at the next signal.

 

[Taunton]

In the area I drive in, mainly 4 aspect, there seems to always be a system in place whereby a conflicting movement is double protected by two reds. So you'll get 2Y, 1Y, R. The red will clear to a 1Y on approach thus having slowed you right down, then you finaly stop at the red actually protecting the movement. Question... is this a national thing or location specific? As annoying as it can be, it does a very good job of reducing the speed on approach to such conflicts and I would have thought especially at a location like this this would be the method of signalling?

This goes right back to manual signalling days, for short sections/short overlaps, known as "Section Clear Junction Blocked", with special bell codes (3-5-5) etc between two adjacent signalboxes. Laverstock North Junction didn't exist in manual days, when it would have had its own signalbox of course, so we can't look back to see how it might have been done then, but it likely would have been done this way.

 

[30907]

YY means double yellow.

But your comment does make one think, what if SY29R had been single yellow because of a movement off the Laverstock curve. I don't believe it has a further outer distant.

Sorry, didn't express myself clearly.
Continuous braking from 29R, whether at Y or YY, would bring the speed down to 0 not 50.
There is no outer distant as the line is 2-aspect up to there.

 

[IanXC]

A simple derailing trap wouldn't be acceptable but a short 'overlap spur' leading to a friction bufferstop or in the old days a sand drag might be an option. Normally these are only provided where there is no room for any kind of sensible clear overlap, yet flexibility demands the ability to run towards the signal while the junction ahead is blocked. I would suggest treating the existing overlap as 'restricted', with an additional 'warning' class route from SY29 to SY31, enforcing a delayed yellow clearance at SY29. The unrestricted 'main' route would also still be available but would have an extended overlap through the junction. Whether the warner could have reduced speed further back sufficient to prevent the slide past, or limit its extent enough to avoid the collision, I really couldn't say. I want to reiterate this is a 40-year-old layout in this form. If it was inherently dangerous, surely that would have been uncovered before. Have there been any worrying SPADs at this junction in the past, or any near misses?

As an example, for those not familiar with this set up, Hambleton North Junction has this set up, presumably so that a service coming down the chord, if it were to pass the signal at danger is not thrust into the path of an up train at 125mph.

Photo pinched from this thread:

https://twitter.com/GarethDennis/status/1232749441121124352/photo/1

 

[MarkyT]

As an example, for those not familiar with this set up, Hambleton North Junction has this set up, presumably so that a service coming down the chord, if it were to pass the signal at danger is not thrust into the path of an up train at 125mph.

Photo pinched from this thread:

https://twitter.com/GarethDennis/status/1232749441121124352/photo/1

Good example, and the arrangement allows the signal to be as close to the junction as possible, so a complete train of the longest formation (probably a freight) can be held comfortably on the chord clear of the junction on the Selby line. Standage is about 530m measured between the signals on Google Earth. For trains going towards Selby, the signal at the other end of the chord has a ~160m overlap clear of the junction, without a trap. The chord and its connections at both ends have a speed limit of 40mph throughout. Interestingly the track and signalling layout is a near contemporary of the Salisbury area. The Selby diversion opened in 1983.

 

[IanXC]

Good example, and the arrangement allows the signal to be as close to the junction as possible, so a complete train of the longest formation (probably a freight) can be held comfortably on the chord clear of the junction on the Selby line. Standage is about 530m measured between the signals on Google Earth. For trains going towards Selby, the signal at the other end of the chord has a ~160m overlap clear of the junction, without a trap. The chord and its connections at both ends have a speed limit of 40mph throughout. Interestingly the track and signalling layout is a near contemporary of the Salisbury area. The Selby diversion opened in 1983.
View attachment 105345

I suppose the principle risk exiting towards Selby would be coming into contact with a train going in the same direction (as per Salisbury in fact), whereas going towards York you could be straight into the path of something at 125 on the Up Main.

In some ways that makes the Salisbury incident more like the hypothetical risk exiting towards Selby. Of course that all changes if that 3rd train had been in the mix at Salisbury.

Interesting its of similar vintage, I hadn't realised that when I thought of it.

 

[Ianno87]

I suppose the principle risk exiting towards Selby would be coming into contact with a train going in the same direction (as per Salisbury in fact), whereas going towards York you could be straight into the path of something at 125 on the Up Main.

Also going towards Selby appears to be a climbing gradient, with a falling gradient towards the ECML.

 

[Watershed]

Good example, and the arrangement allows the signal to be as close to the junction as possible, so a complete train of the longest formation (probably a freight) can be held comfortably on the chord clear of the junction on the Selby line. Standage is about 530m measured between the signals on Google Earth. For trains going towards Selby, the signal at the other end of the chord has a ~160m overlap clear of the junction, without a trap. The chord and its connections at both ends have a speed limit of 40mph throughout. Interestingly the track and signalling layout is a near contemporary of the Salisbury area. The Selby diversion opened in 1983.
View attachment 105345

An overlap spur coming off the Down line, towards the 'V' of Salisbury Tunnel Jn, would still present the risk that an errant train overruns the buffers and collides with a train on the Romsey lines.

At a converging junction there is no easy way of preventing such a collision, short of having something like double red protection.

 

[snookertam]

A lot of interesting replies and suggestions. Could I venture one? Straight forward SPAD mitigation which is commonly used at certain junctions around Glasgow (and I’m sure elsewhere…). Basically where approaching a junction where a conflicting move is taking place, the train is stopped at the second signal in the rear of the junction, where that which protects the junction is quite close to the junction itself. It’s still regularly done around Cathcart, and certainly used to be done on the approach to Cowlairs from Bishopbriggs.

I believe it was introduced in the aftermath of the Bellgrove and Newton head on collisions at single lead junctions. It was introduced at a time where there was no TPWS of course, which may have been an influence in the preference for the approach. It would also only be practical in areas where the service level would allow it, or else it would cause additional congestion. But as a policy, perhaps at least during autumn, it could be a worthwhile fail safe. As we have seen here, despite systems like TPWS, if the brakes aren’t going to perform there’s little that can be done. It would be better to find that out when the train isn’t approaching a junction.

A very simple measure that costs nothing, but hindsight and all that…

 

[Watershed]

A very simple measure that costs nothing

Apart from requiring a complete rewrite of the timetable and increasing journey times, yes...

 

[snookertam]

Apart from requiring a complete rewrite of the timetable and increasing journey times, yes...

As I said above, it would only be workable in areas where the service level allowed it. In this scenario the GWR train was significantly late, so if introduced it would have had no effect on the timetable itself. If the GWR service had been on time the SWR service wouldn’t have been cautioned/halted by signals.

 

[Bald Rick]

A very simple measure that costs nothing, but hindsight and all that…

It does cost something - in the changes required to the interlocking.

 

[30907]

Another possibility, which also doesn't come cheap, is to move SY29 further out (perhaps onto the straight section about 200m out); but if 29R is correctly positioned you would then have to move that as well.....

All this assumes that - contrary to what SWR have stated - the driver was at fault in not responding correctly to signals 29R and 29.