Incident at Talerddig, Wales - 21/10/2024

[Lurcheroo]

It seems that after normal braking was insufficient to slow the train enough emergency braking was applied until the collision. I seem to recall that repeated braking may be more efficient in slides that continuous braking. If the ERTMS can override braking (reading a comment above) and apply the emergency brake does this mean that the driver cannot override this? If the rail adhesion was too low to enable the train wheels to adhere then I assume any type of braking would not stop the train particularly as sanding was not working. Perhaps also the driver applied emergency brake and then left the cab when the collision was unavoidable.

On a 158 a driver can not override any emergency brake demand, Including one from the ERTMS system.

I don't know what TfW driver instructions are, but for us with WSP fitted stock if you encounter a full slide you must not attempt to regain rotation, you just stick it in emergency and wait until it stops.

Yep, same, let WSP do it’s job, do not release the brake, select a higher level of braking if required.

The question might also be legitimately asked why 1980s generation stock hasn't been retro-fitted with new WSP equipment in the last 20 years or so, in the same way that all the HST operators (except FGW/GWR and ScotRail) have done?

There is mention online that there was something done to the WSP when TFW had them refurbished, what exactly I don’t know.

One point I’m interested to clarify - it has been stated that the RHTT ran approx 24hrs before the accident. However it’s also been said that ERTMS will always route the first train to arrive at Talerddig into the loop, so that the second train can pass on what is nominally now the ’through line.’

A minor clarification incase it’s relevant to your thinking, the route isn’t set by the ERTMS system, the signaller sets the route, but they should always set the route for the first train into the loop.
At Newtown and Welshpool, the RHTT does one side in each direction.

 

[starlight73]

I got myself confused so I’ve found the location on Open Railway Map: https://www.openrailwaymap.org/?style=standard&lat=52.58877520751781&lon=-3.5786890983581543&zoom=15

As others have said, the collision happened west of the passing loop at Talerddig and the refuge siding is at the eastern end, visible if you zoom in.

 

[Whistler40145]

As a matter of interest, how often would any maintenance work be carried out on the Class 158 Sanding Equipment to ensure it's fully functional?

 

[Taunton]

I always thought Talerddig was at the pinnacle of a substantial climb from both sides, so can't quite get the train braking on the uphill approach and going into a slide there before even taking the loop.

 

[Snow1964]

I always thought Talerddig was at the pinnacle of a substantial climb from both sides, so can't quite get the train braking on the uphill approach and going into a slide there before even taking the loop.

From Yesterdays announcement

Westbound trains climb an ascending gradient to enter the loop and, on exiting, rejoin the single track as it descends towards Llanbrynmair and Machynlleth.

Train 1J25 then entered the loop at Talerddig. Although the train slowed while passing through the loop, it did not stop before passing the block marker positioned near the exit. The train subsequently exited the loop, rejoined the single line, and continued to travel for around 900 metres on the descending gradient, before colliding with train 1S71.

Collision between passenger trains near Talerddig

Investigation into a collision between passenger trains near Talerddig, Powys, on 21 October 2024.

www.gov.uk

The question of train speed (at point of collision) is still subject to investigation, but at this stage can't rule out it speeding up again on the downhill section, as it managed to go extra 900m before colliding.

 

[edwin_m]

From Yesterdays announcement

Collision between passenger trains near Talerddig

Investigation into a collision between passenger trains near Talerddig, Powys, on 21 October 2024.

www.gov.uk

The question of train speed (at point of collision) is still subject to investigation, but at this stage can't rule out it speeding up again on the downhill section, as it managed to go extra 900m before colliding.

This does mean that the train probably wasn’t braking much or at all on the approach to the loop, because of the ascending gradient. Wheelspin in traction might have alerted the driver to poor adhesion on the ascent, but if the rail was clean or the sanders were working on previous brake applications the driver might not have been aware of the inability to decelerate until it was too late. As soon as it slid past the end of the loop and onto the steep descent, it would have been unable to slow down until the gradient eased or the adhesion improved.

 

[12LDA28C]

As a matter of interest, how often would any maintenance work be carried out on the Class 158 Sanding Equipment to ensure it's fully functional?

As previously mentioned in post #188, the sanders should be tested as part of a train prep which should be done before the unit enters service each day.

 

[big bus driver]

I think no autumn driving policy was going to prevent this (I appreciate that isn’t what your trying to say), 40 seconds of braking before an emergency braking application was initiated, which if it wasn’t by the driver, the ERTMS system would have done so as soon as it determined that there was insufficient distance to stop the train before the block marker based on the current speed.

Thanks, i tend to agree that no TOCs defensive driving policy could probably have prvented this. I was more wondering where TFW are with briefing and training, as I’m aware some TOCs have removed ‘skid pan’ training and off train autumn briefs to save money, where others still consider them valuable.

This years leaf fall season is reminiscent of how it always used to be 10/15 years ago before WSP and the RHTT. I’d hazard a guess well over half of the drivers on the mainline have been driving less time than this, so will have no experience of how to drive in these conditions, other than what they get via training.

 

[Llanigraham]

At Newtown and Welshpool, the RHTT does one side in each direction.

It does the same at Talerddig loop.

 

[InkyScrolls]

I always thought Talerddig was at the pinnacle of a substantial climb from both sides, so can't quite get the train braking on the uphill approach and going into a slide there before even taking the loop.

You underestimate a train's momentum!

 

[Siemens Staines]

IF I have traced it back correctly, on the day in question 158841 started from Chester P3B vice Crewe CS, so the unit prep would need to have been carried out at a different location to that which it is normally booked.

Realtime Trains | 5D03 0415½ Chester to Shrewsbury | 21/10/2024

Real-time train running information for 5D03 0415½ departure from Chester to Shrewsbury on 21/10/2024. From Realtime Trains, an independent source of train running info for Great Britain.

www.realtimetrains.co.uk

 

[Belperpete]

Forgive the minor divergence from the subject, but still on topic I suppose..

When ERTMS becomes aware a block marker is likely to be overshot or is overshot, does it do anything to trains on conflicting routes?
As in would it put the emergency brakes on the 'non offending' train to reduce the collision forces?

I can't answer this definitively, but I would doubt it. The ERTMS consists of two systems, the train-borne equipment and the central equipment connected to the signalling interlocking. The train-borne equipment will know that it is likely to overshoot its stopping point, and will apply the emergency brake. The assumption then is that the train will be stopped within the overlap, clear of other trains. The train bourne equipment is unlikely to know if it is going to come into conflict with another train, or even what other trains are in the vicinity.

Most interlockings will detect when a train over runs a signal, or stop marker, and will alert the signaller. In modern interlockings, it will also automatically replace signals on conflicting routes. What I don't know is what the particular interlocking used at Mach does, or what ERTMS does equivalent to replacing signals. The last time I was involved with ERTMS, there was a big debate about what to do about a train that has already passed the stop marker. If you automatically stop a train that is close to a junction when another train over runs, you could end up stopping that train in the path of the over running train, when otherwise it might have got safely clear.

The report states that the train coming from Mach was only travelling at about 5mph, which suggests that it was already braking. This could be because its driver could see the train heading towards him, or because the signaller had been alerted to the SPAD and had sent out an emergency stop message, or because the interlocking had automatically triggered a stop message. It is also possible that the driver of the SPADing train could have used the train radio to raise the alarm. No doubt the full report will give more details.

 

[Lurcheroo]

I always thought Talerddig was at the pinnacle of a substantial climb from both sides, so can't quite get the train braking on the uphill approach and going into a slide there before even taking the loop.

It is, the line speed towards Tallerdig from Shrewsbury is 130KPH then drops to 95kph, if going over the points it then drops to 50kph.

== Doublepost prevention - post automatically merged: 6 Nov 2024 ==

IF I have traced it back correctly, on the day in question 158841 started from Chester P3B vice Crewe CS, so the unit prep would need to have been carried out at a different location to that which it is normally booked.

Realtime Trains | 5D03 0415½ Chester to Shrewsbury | 21/10/2024

Real-time train running information for 5D03 0415½ departure from Chester to Shrewsbury on 21/10/2024. From Realtime Trains, an independent source of train running info for Great Britain.

www.realtimetrains.co.uk

Interesting! Might be rather difficult for anyone preparing it for service to do any checks below the solebar if it’s stabled at a platform!

 

[millemille]

As previously mentioned in post #188, the sanders should be tested as part of a train prep which should be done before the unit enters service each day.

Train prep and/or fuel point exams doesn't ensure that a sander is fully functional. It's a visual check to confirm that sand is seen coming out of the hoses but doesn't check the actual amount of sand coming out.

Most fleets have an annual "autumnisation" campaign that runs leading up to leaf fall where the sanders are subject to full testing where the weight of sand expelled for a timed period is measured to ensure the correct delivery rate.

Things like sand hose visual checks for damage and blocking will be on every exam, so approximately once a month.

 

[12LDA28C]

This years leaf fall season is reminiscent of how it always used to be 10/15 years ago before WSP and the RHTT. I’d hazard a guess well over half of the drivers on the mainline have been driving less time than this, so will have no experience of how to drive in these conditions, other than what they get via training.

What do you mean by 'this years leaf fall season'? Do you have any figures to demonstrate that this year is any worse than previous years, as that seems to be a sweeping generalisation to make without any evidence. Also, WSP has been around a lot longer than '10/15 years'.

== Doublepost prevention - post automatically merged: 6 Nov 2024 ==

The report states that the train coming from Mach was only travelling at about 5mph, which suggests that it was already braking. This could be because its driver could see the train heading towards him, or because the signaller had been alerted to the SPAD and had sent out an emergency stop message, or because the interlocking had automatically triggered a stop message. It is also possible that the driver of the SPADing train could have used the train radio to raise the alarm. No doubt the full report will give more details.

This was the case

== Doublepost prevention - post automatically merged: 6 Nov 2024 ==

It is also possible that the driver of the SPADing train could have used the train radio to raise the alarm. No doubt the full report will give more details.

The REC wasn't made until the collision had occurred.

 

[43096]

The REC wasn't made until the collision had occurred.

Given that the driver must have been expecting to pass something, if that is correct then it is surprising - as soon as the driver knew he was going to enter the single line beyond the loop you’d expect him to initiate a REC.

 

[Bald Rick]

This does mean that the train probably wasn’t braking much or at all on the approach to the loop, because of the ascending gradient.

The RAIB update (quoted in full above) says the driver applied service braking for 40 seconds, and wheel slide occurred in that period; then a Emergency brake demand occurred (RAIB does not say whether this was by the driver or ETCS intervening) and then the train entered the loop.

My reading of that is that the train was braking for at least 40 seconds on the approach to the loop. At usual service braking rates that would reduce speed between 55 and 85 km/h. How much it was actually decelerating is a different question!

 

[12LDA28C]

My reading of that is that the train was braking for at least 40 seconds on the approach to the loop. How much it was decelerating is a different question!

Indeed although one might assume that if the train speed was not decelerating sufficiently in line with the driver's expectations, the brake might have been placed to full service or emergency rather before 40 seconds had elapsed. The report mentions 'service braking' so no detail of the actual brake step(s) involved.

 

[357]

What do you mean by 'this years leaf fall season'? Do you have any figures to demonstrate that this year is any worse than previous years, as that seems to be a sweeping generalisation to make without any evidence. Also, WSP has been around a lot longer than '10/15 years'.

I agree with you, but I will say that from my own experience this year has been the worst since I've been driving.

 

[Killingworth]

The RAIB update (quoted in full above) says the driver applied service braking for 40 seconds, and wheel slide occurred in that period; then a Emergency brake demand occurred (RAIB does not say whether this was by the driver or ETCS intervening) and then the train entered the loop.

My reading of that is that the train was braking for at least 40 seconds on the approach to the loop. How much it was decelerating is a different question!

The RAIB wil be calculating as accurately as possible the speed of the train on it's approach and the precise points at which normal and emergency brakes were applied. It's very clear that it slid a very long way, as did the units at Salisbury. The rail and tree conditions in that detailed area are very relevant.

That's all self evident. In the meantime those involved in the incident must be even more interested in what could have happened differently and will welcome the end of speculation when the RAIB reports.

 

[Belperpete]

The RAIB update (quoted in full above) says the driver applied service braking for 40 seconds, and wheel slide occurred in that period; then a Emergency brake demand occurred (RAIB does not say whether this was by the driver or ETCS intervening) and then the train entered the loop.

Agreed that it starts a new para by saying "Train 1J25 then entered the loop at Talerddig." However, it isn't actually clear what bit of the preceding paragraph the "then" is referring to. If read literally it says that the train then entered the loop after the events covered by the previous para (i.e. after the collision). Hopefully the final report will be a bit more precisely written.

 

[irish_rail]

IF I have traced it back correctly, on the day in question 158841 started from Chester P3B vice Crewe CS, so the unit prep would need to have been carried out at a different location to that which it is normally booked.

Realtime Trains | 5D03 0415½ Chester to Shrewsbury | 21/10/2024

Real-time train running information for 5D03 0415½ departure from Chester to Shrewsbury on 21/10/2024. From Realtime Trains, an independent source of train running info for Great Britain.

www.realtimetrains.co.uk

If the train was prepped in Chester Station at a platform then it is highly unlikely the driver prepping it would have been able to test the sanders, as this is done at track level (obviously impossible when train is at a platform).

 

[BrummieBobby]

From my experience signalling, I cannot add anything further from a technical point of view. However just to say that the more I read, I am filled with pity for the driver of 1J25; to apply the service brake as per normal driving technique, then for the emergency to go in and still slide for over a mile, with signals against the train, knowing that he was sliding into a single line section with a collision likely.

Just awful.

 

[DelW]

From my experience signalling, I cannot add anything further from a technical point of view. However just to say that the more I read, I am filled with pity for the driver of 1J25; to apply the service brake as per normal driving technique, then for the emergency to go in and still slide for over a mile, with signals against the train, knowing that he was sliding into a single line section with a collision likely.

Just awful.

It was reported (link from post 143 above) that the driver had time to get back into the saloon and warn passengers of the likely impending collision. Presumably there was nothing further he could do in the cab at that stage.

 

[GC class B1]

It seems that after normal braking was insufficient to slow the train enough emergency braking was applied until the collision. I seem to recall that repeated braking may be more efficient in slides that continuous braking. If the ERTMS can override braking (reading a comment above) and apply the emergency brake does this mean that the driver cannot override this? If the rail adhesion was too low to enable the train wheels to adhere then I assume any type of braking would not stop the train particularly as sanding was not working. Perhaps also the driver applied emergency brake and then left the cab when the collision was unavoidable.

My understanding is that there is no functional difference between Full Service and Emergency braking rates on class 158 units.

 

[CC 72100]

My understanding is that there is no functional difference between Full Service and Emergency braking rates on class 158 units.

Correct, in terms of braking force, Step 3 and Emergency produce the same force.

 

[12LDA28C]

Correct, in terms of braking force, Step 3 and Emergency produce the same force.

So what is the point of having an 'emergency' position of the brake then, if if provides no extra brake force? Might as well not be there.

 

[bahnause]

So what is the point of having an 'emergency' position of the brake then, if if provides no extra brake force? Might as well not be there.

Could be a fail-safe channel for activating the brake.

 

[bramling]

So what is the point of having an 'emergency' position of the brake then, if if provides no extra brake force? Might as well not be there.

The emergency position should be hard-wired in such a way as to be fail-safe.

 

[InkyScrolls]

So what is the point of having an 'emergency' position of the brake then, if if provides no extra brake force? Might as well not be there.

It earths the braking system to avoid any chance of the brake being released by mistake until it has timed out.