Incident at Talerddig, Wales - 21/10/2024

[43096]

There's two ways to look at this really, for example in the case of Salisbury, once the driver knew they were going to pass the signal, an earlier REC (I must admit I can't recall if there was one or not, I try not to remember that night at work!) could have got the driver of the GWR to stop in a potentially more vulnerable place than it did.

Yes, agreed on that. Same also applies to the sleeper brake failure incident at Edinburgh.

But in this case, on a single line where the driver is expecting to pass a train travelling the other way, immediate initiation of a REC has to be the best thing to do.

 

[norbitonflyer]

There's two ways to look at this really, for example in the case of Salisbury, once the driver knew they were going to pass the signal, an earlier REC (I must admit I can't recall if there was one or not, I try not to remember that night at work!) could have got the driver of the GWR to stop in a potentially more vulnerable place than it did.

Indeed - a third train could have been involved as the Salisbury to Southampton train had only recently cleared the junction when the collision between the other two happaned.

 

[Lurcheroo]

Points are designed to be run through without (much) risk of derailment. For traditional point machines and mechanical installations, that usually means bent or broken thrust, lock and detection rods. In clamp locks, the force on switch rails overcomes the clamping force. For the HPSS, manufacturers had to incorporate a 'frangible link' to get approval. In all cases, the point mechanism is broken in some way after a run through and needs repair before being placed back in service.

I have heard that if the points were a different type, on this occasion, it is expected that it would have derailed the train.

 

[MarkyT]

I have heard that if the points were a different type, on this occasion, it is expected that it would have derailed the train.

Going by the loop points at Dovey Junction, I expect the Talerrdig ones are also clamp locks now, probably the preferred in-bearer type (has the hydraulic rams and much of the mechanism inside a big steel box under the rails which replaces a bearer (sleeper) and manages the cabling). They won't be the old hydropneumatic self-acting ones of RETB days anymore. I don't know of any point mechanisms that can't be trailed. The HPSS pilot installation 20+ years ago didn't have the frangible link that would fail in a run-through I think, but it was insisted on for wider network approval and the pilot ones were updated.

 

[Lurcheroo]

Going by the loop points at Dovey Junction, I expect the Talerrdig ones are also clamp locks now, probably the preferred in-bearer type (has the hydraulic rams and much of the mechanism inside a big steel box under the rails which replaces a bearer (sleeper) and manages the cabling). They won't be the old hydropneumatic self-acting ones of RETB days anymore. I don't know of any point mechanisms that can't be trailed. The HPSS pilot installation 20+ years ago didn't have the frangible link that would fail in a run-through I think, but it was insisted on for wider network approval and the pilot ones were updated.

They look like clamp locks to me, definitely not hydropneumatic Now.

 

[Belperpete]

They look like clamp locks to me, definitely not hydropneumatic Now.

Agreed. All the hydro-pneumatics installed when the line was signalled with RETB were replaced when the line was resignalled with ETCS/ERTMS.

 

[Annetts key]

Are they sprung, or locked in position?

All modern points that have facing moves on a passenger line have to have a facing point lock (or equivalent system).

I have heard that if the points were a different type, on this occasion, it is expected that it would have derailed the train.

Points are designed to be run through without (much) risk of derailment. For traditional point machines and mechanical installations, that usually means bent or broken thrust, lock and detection rods. In clamp locks, the force on switch rails overcomes the clamping force. For the HPSS, manufacturers had to incorporate a 'frangible link' to get approval. In all cases, the point mechanism is broken in some way after a run through and needs repair before being placed back in service.

Going by the loop points at Dovey Junction, I expect the Talerrdig ones are also clamp locks now, probably the preferred in-bearer type (has the hydraulic rams and much of the mechanism inside a big steel box under the rails which replaces a bearer (sleeper) and manages the cabling). They won't be the old hydropneumatic self-acting ones of RETB days anymore. I don't know of any point mechanisms that can't be trailed. The HPSS pilot installation 20+ years ago didn't have the frangible link that would fail in a run-through I think, but it was insisted on for wider network approval and the pilot ones were updated.

I've attended far more run throughs (where a train has passed over a trailing point that was not set for the movement that occurred) than I care to remember. And in every case, the train, machine (e.g. tamper) or on track plant (road/rail "360" like a JCB) never derailed. I'm not saying will never happen but it is very unlikely to occur.

As well as the substantial damage to the point operating mechanism, many of all of the stretcher/tie bars will be damaged (broken or bent), if a conventional point machine, the FPL stretcher bar will be bent. Because the point operating equipment tried to prevent the closed switch rail from moving, the top inside edge of the switch rail will have been bent outwards at the tip. Plus the whole front part may have rotated slightly outward so that it now will no longer full close up to the fixed rail at the top.

Before the 'frangible link' was included as part of HPSS, it was said that it was possible for a train/machine to become trapped by the switch rail gripping the wheel. I never saw this myself.

 

[GC class B1]

So the same as on other traction types then in that it can't be released until the train has come to a stand. But not the same as on other traction types as it provides no additional brake force (20% extra on a Class 165, as an example).

The UIC brake on locomotive hauled trains has the same retardation rate in Full Service and Emergency and does not have a timer for a driver applied Emergency brake application and as a result an Emergency brake application can be released once the brake pipe has been recharged. The only difference between Full Service and Emergency applications with this system is that the Emergency application time is quicker.

 

[Killingworth]

Sanding. As a layman I'm curious as to how much effect this could have had on a sliding train with brakes apparently locked going downhill, assuming the sanding unit had worked. With a 2 car unit how many wheels would benefit from sand and potentially for how long?

 

[12LDA28C]

The UIC brake on locomotive hauled trains has the same retardation rate in Full Service and Emergency and does not have a timer for a driver applied Emergency brake application and as a result an Emergency brake application can be released once the brake pipe has been recharged. The only difference between Full Service and Emergency applications with this system is that the Emergency application time is quicker.

I'm sure you're correct, but we are not talking about loco-hauled trains.

 

[sharpener]

As well as the substantial damage to the point operating mechanism, many of all of the stretcher/tie bars will be damaged (broken or bent), if a conventional point machine, the FPL stretcher bar will be bent. Because the point operating equipment tried to prevent the closed switch rail from moving, the top inside edge of the switch rail will have been bent outwards at the tip. Plus the whole front part may have rotated slightly outward so that it now will no longer full close up to the fixed rail at the top.

It sound like a lot of work to replace/re-adjust the switch rails and all the various bars involved.

Asking as another layman with some experience of machinery design, is the purpose of the mandatory frangible link just to ensure the point blades will not trap the wheels in a run-through situation?

If it is intended to prevent/mitigate damage it does not sound as though it is adequate. I would have thought it would be possible to design a point mechanism (with or without one or more frangible links) that would lock against facing moves but still be able to avoid what sounds like pretty major damage when forced over in the trailing direction.

 

[norbitonflyer]

There's two ways to look at this really, for example in the case of Salisbury, once the driver knew they were going to pass the signal, an earlier REC (I must admit I can't recall if there was one or not, I try not to remember that night at work!) could have got the driver of the GWR to stop in a potentially more vulnerable place than it did.

The Talerdigg collision happened right next to a trunk road, which made the emergency service's job much easier than it might have been. Either fortuitous, or some quick thinking on the part of the driver of the eastbound train to bring it to a halt where it did.

Edinburgh has been cited as another example where stopping everything could have had worse consequences than letting everything get out of the way. The Northern Line runaway grinder a few years back is another example.

 

[InkyScrolls]

Sanding. As a layman I'm curious as to how much effect this could have had on a sliding train with brakes apparently locked going downhill, assuming the sanding unit had worked. With a 2 car unit how many wheels would benefit from sand and potentially for how long?

Only the third axle in the direction of travel receives sand on a 158 (as built - some have mods, not sure if the TfW 158s do).

 

[edwin_m]

Sanding. As a layman I'm curious as to how much effect this could have had on a sliding train with brakes apparently locked going downhill, assuming the sanding unit had worked. With a 2 car unit how many wheels would benefit from sand and potentially for how long?

When braking in poor adhesion the WSP should release and re-apply the brakes as soon as it detects a locked axle, which among other things should allow the sand to get to where it is needed. I was involved with the one-shot sander originally fitted to some 158s in the 1990s, including seeing analysis of on-train recording data showing little or no deceleration on a poor rail becoming close to normal braking as soon as sand was applied.

Only the third axle in the direction of travel receives sand on a 158 (as built - some have mods, not sure if the TfW 158s do).

The 158s as built did not have sanders. The first fitment was the one-shot version mentioned above on 159s and some 158s, activated manually when needed and dispensing ahead of the first axle. I believe all 158s now have a version that can dispense sand multiple times and triggered automatically. These do indeed work on the third axle, partly due to concerns about the sand preventing track circuit actuation. The effect carries over to axles further back in a shorter train, but longer trains have less problem with adhesion because the wheels nearer the front tend to clean the rails so the ones further back get a better grip without the use of sand.

It sound like a lot of work to replace/re-adjust the switch rails and all the various bars involved.

Asking as another layman with some experience of machinery design, is the purpose of the mandatory frangible link just to ensure the point blades will not trap the wheels in a run-through situation?

If it is intended to prevent/mitigate damage it does not sound as though it is adequate. I would have thought it would be possible to design a point mechanism (with or without one or more frangible links) that would lock against facing moves but still be able to avoid what sounds like pretty major damage when forced over in the trailing direction.

I'd guess the frangible link in the HPSS would be designed so it brakes in the event of being trailed through before any other damage occurs, so repair would be a matter of replacing the link. Other systems that don't have that feature might be the ones that get damaged more severely.

 

[Belperpete]

It sound like a lot of work to replace/re-adjust the switch rails and all the various bars involved.

Asking as another layman with some experience of machinery design, is the purpose of the mandatory frangible link just to ensure the point blades will not trap the wheels in a run-through situation?

If it is intended to prevent/mitigate damage it does not sound as though it is adequate. I would have thought it would be possible to design a point mechanism (with or without one or more frangible links) that would lock against facing moves but still be able to avoid what sounds like pretty major damage when forced over in the trailing direction.

It is a lot of work to fix a run-through. In my experience, you are into replacing the stretcher bars and connecting rods, and probably the point machine too, or clamps in the case of a clamp-lock.

It is easy to say that something should be possible. I would have thought that if it were possible, someone would have come up with a solution by now. I vaguely recall that there was a version of the clamp-lock that could be trailed, but it was only for use in yards as it wasn't considered to adequately lock them for facing passenger moves. Likewise there are some continental designs that unlock a set of points if a train trails through, but AFAIA they aren't used in the UK, again because they are not considered to adequately lock the points to UK requirements.

Remember that both switch blades are physically locked in position - it is as important that the open blade stays open as the closed blade stays closed. They have to stay locked not just against the forces applied by the train wheels at the switch tips, but against the forces applied on the blades as the train passes along the switch blade - these forces can try to open the closed switch tip. The two switch blades are physically connected not just at the switch tips, but by stretcher bars further along the switches too, to prevent the open switch from flexing and coming into contact with the backs of the wheels. So as a train takes a point in the facing direction, there can be forces on both the open and closed blades, similar to those that would be applied by a train trailing through the points. So anything that unlocks the switches due to the forces applied by a train trailing through the points could also unlock them due to the indirect forces applied by a train taking the points in the facing direction.

 

[Harpo]

I would have thought it would be possible to design a point mechanism (with or without one or more frangible links) that would lock against facing moves but still be able to avoid what sounds like pretty major damage when forced over in the trailing direction.

Needing a piece of kit that over-rides all of the other kit that’s locking the point in a facing position and providing a safe wheeled route in the intended direction with all of its assurances in the interlocking. That in turn would need (expensive!) safeguards to ensure that it didn’t unintentionally release.

Probably cheaper (and safer) to replace occasional run-through damage.

 

[Annetts key]

In theory, during the normal operation of the railway, no train, machine or road/rail plant should ever run-through power operated or mechanical operated points controlled by a signal box / ground frame. If normal signalling is in use and the rules are followed, the points should always be set for the correct position before the movement can be authorised.

The vast majority of run-through incidents occur during T3 engineering occupations because either the PICOP or Engineering Supervisor failed to ensure that the points were set correctly for the movement. A driver of a train/machine/plant is also supposed to check that the points are set correctly before passing over them.

The other occasions are when a train passes a signal or block marker that it should have stopped at or when there is a signalling failure or other problem and the signaller is verbally authorising the movement. The route setting agent has the responsibility to ensure all the points are set to the correct position. But very occasionally a mistake is made.

The point operating equipment has no way of knowing which direction the train is approaching. The interlocking could be designed to determine the direction at increased cost. however this is only possible if normal signalling is in use. Providing all the extra equipment to unlock power operated points would increase the complexity and cost even more money that the railway does not have.

Hence I don't think it likely that you will see any changes in this regard on the Network Rail network.

The power operated points that did not need to be set for trailing moves that I worked on were not clamp locks but did use a clamp lock hydraulic power pack and clamp lock hydraulic rams. It's important to know that a clamp lock hydraulic power pack has an over pressure release value (I think it operates at 90 bar) that allows the oil back into the tank. There was no FPL provided.This is the reason they were only used on non-passenger lines.

A conventional electrical detector was used to detect the position of the switch rails. In addition two hydraulic pressure sensor switches were used. IIRC one set at 60 bar and one set at 50 bar. If the 60 bar switch changed state, the interlocking powered up the power pack to restore the pressure and hence ensured that the switch rail returned or stayed in the commanded position. If the 50 bar switch changed state, the this dropped out the detection and hence put the signals back to red.

Also of note was that the interlocking did power the points to the correct position for trailing moves if the signalling was being used.

The installation I described was at Bristol SPM HST Depot (as it was known at the time). At first it was very reliable. But only two of the boxes that contained the hydraulic pressure sensor switches were supplied. One was installed with one spare. Eventually the installed unit failed and was replaced with the spare. Then that eventually failed. No further spares were available. For many years it operated in a degraded mode (the drivers reporting the signals flashing from off to on and then back to off). Then it was replaced by a conventional clamp lock point.

 

[Dunfanaghy Rd]

If it is intended to prevent/mitigate damage it does not sound as though it is adequate. I would have thought it would be possible to design a point mechanism (with or without one or more frangible links) that would lock against facing moves but still be able to avoid what sounds like pretty major damage when forced over in the trailing direction.

There is. Used all over the Continent, often called a 'Claw Lock'. OS Nock was an advocate for its adoption in UK. It seems (to me) to fall foul of an 1840 requirement to have 2 (minimum) fixed length stretchers. The Australian Rail Track Corp. used to have some stuff on them that I downloaded 14 years ago.
Pat

 

[tumbledown]

This?

== Doublepost prevention - post automatically merged: 8 Nov 2024 ==

There is. Used all over the Continent, often called a 'Claw Lock'. OS Nock was an advocate for its adoption in UK. It seems (to me) to fall foul of an 1840 requirement to have 2 (minimum) fixed length stretchers. The Australian Rail Track Corp. used to have some stuff on them that I downloaded 14 years ago.
Pat

 

[Belperpete]

There is. Used all over the Continent, often called a 'Claw Lock'. OS Nock was an advocate for its adoption in UK. It seems (to me) to fall foul of an 1840 requirement to have 2 (minimum) fixed length stretchers. The Australian Rail Track Corp. used to have some stuff on them that I downloaded 14 years ago.
Pat

As you say, UK requirements require the two switch rails to be coupled together with at least two stretcher bars, in case one fails. The claw lock will not work if you do that:

There are no fixed gauge rods (stretchers, front rods, backs rods) connecting the switches since the switches must be able to move independently of one another at each end of the drive stroke to allow the claws to unlock and lock.

With the claw-lock, the two switches are held by a single coupling bar between the two claws, which are connected to the rails. If this single coupling bar should fail, or the claw comes off the rail, one of the switch rails could move. I presume that this is why it was never accepted for UK use.

 

[Annetts key]

Nowadays it's unusual to find new installations that have less than three stretchers on a passenger line. Plus points are almost always provided with a back drive. This needs all the stretchers in order for it to work.

But this is all going rather off topic. The point being run through made no difference to the events that occurred.

 

[Harpo]

But this is all going rather off topic. The point being run through made no difference to the events that occurred.

Quite. The absence of flank protection determines the outcome. A lesser version of SN109’s flaws?

 

[Signal Head]

I can't speak about modern brake control systems, but with traditional air brakes, whereas going to "Full Service" position bleeds the air out in a controlled manner down to (from memory ) about 48psi (so the brakes are applied gently), "Emergency" opens the Train Pipe 'full bore', directly to atmosphere so the pressure is completely vented and the brakes apply quicker.

 

[Annetts key]

Quite. The absence of flank protection determines the outcome. A lesser version of SN109’s flaws?

We get back to the principle that only braking systems that are used correctly (driver applying braking in good time) and which are effective (increase the friction enough) stop trains.

This is different to the Ladbroke Grove, Southall and many other incidents where a train passed a signal at danger. In this case it appears the driver did apply the brake and did everything they could to try to stop the train.

There have been other incidents where a train braking system was unable to stop a train in the normal distance. Salisbury has already been mentioned. But there are others (many of which will not be relevant here, as it appears the braking system itself was working correctly).

I'm not sure any substantial changes in infrastructure are going to be recommended. Given how far this train slid, there is no easy infrastructure change that would have prevented some kind of accident. Yes, trap points followed by a sand drag could have helped. But that is an expensive solution and it would cost far too much to do this everywhere. Even providing this at selected "at risk" locations would cost rather a lot of money. Money that could be better spent on a better solution.

 

[Belperpete]

Agreed. While a trap point and sand drag might have avoided the collision, it would likely have resulted in significantly more injuries on the SPADing train. At the speed it was travelling, you would need a significant length of sand drag to stop it safely in a controlled manner.

Flank protection isn't practical or affordable at every conflict. The emphasis has surely got to be to ensure that trains can stop when they are supposed to.

 

[Bikeman78]

Agreed. While a trap point and sand drag might have avoided the collision, it would likely have resulted in significantly more injuries on the SPADing train. At the speed it was travelling, you would need a significant length of sand drag to stop it safely in a controlled manner.

Flank protection isn't practical or affordable at every conflict. The emphasis has surely got to be to ensure that trains can stop when they are supposed to.

Is there any news regarding double variable rate sanders? Modern Railways has mentioned them a few times but progress on fitting them to trains seems to be very slow.

 

[millemille]

Is there any news regarding double variable rate sanders? Modern Railways has mentioned them a few times but progress on fitting them to trains seems to be very slow.

See post #158 in this thread.

But don't conflate DVRS (or SVRS in the theoretical case here) with preventing the accident, if the sand delivery hoses are blocked or restricted the kind of sander is irrelevant. Sanders only work by sand getting into the wheel/rail interface, being crushed and conditioning the rail head.

 

[Bald Rick]

Is there any news regarding double variable rate sanders? Modern Railways has mentioned them a few times but progress on fitting them to trains seems to be very slow.

They are not cheap, and there are as yet no plans to fit any rolling stock other than wht is already fitted / about to be fitted.

 

[ThirteenArches]

While a trap point and sand drag might have avoided the collision, it would likely have resulted in significantly more injuries on the SPADing train. At the speed it was travelling, you would need a significant length of sand drag to stop it safely in a controlled manner.

What you are actually saying here is that this line can not be operated safely under current conditions. A safe method of operation would require one train to be detected as halted in a loop before the crossing train is released into the section to that loop. Looking at timetables from Caersws to Machynlleth, this would require a quarter- hour or so wait for one of the trains, which might well make the service unviable. I suppose an alternative would be a really restrictive speed limit, the descending train literally crawling from the summit into the loop at a rate which would make sliding through vanishingly unlikely. Good old ALARP again.

 

[MrJeeves]

What you are actually saying here is that this line can not be operated safely under current conditions. A safe method of operation would require one train to be detected as halted in a loop before the crossing train is released into the section to that loop. Looking at timetables from Caersws to Machynlleth, this would require a quarter- hour or so wait for one of the trains, which might well make the service unviable. I suppose an alternative would be a really restrictive speed limit, the descending train literally crawling from the summit into the loop at a rate which would make sliding through vanishingly unlikely. Good old ALARP again.

With that logic, basically every line and junction on the entire network is "unsafe" because a train can blast through some points into the path of another train.