• Our new ticketing site is now live! Using either this or the original site (both powered by TrainSplit) helps support the running of the forum with every ticket purchase! Find out more and ask any questions/give us feedback in this thread!

Incident at Talerddig, Wales - 21/10/2024

Lurcheroo

Established Member
Joined
21 Sep 2021
Messages
1,239
Location
Wales
Did the train know that it was sliding though? If individual axles have a difference in speed it can tell, but if the whole lot locks up in one go then the speedo will just drop to zero and the unit will think that it has stopped.

I can't speak for the 158, but more modern WSP systems do things like noticing abnormal deceleration and even releasing the brake on one axle to get a true reading of speed. I think some may even have Doppler radio to measure actual speed over the ground.

as ERTMS uses two separate sources of train speed, that it would not drop to zero even if the WSP did lose control and cease to produce a reliable speed over ground signal
I can confirm that millemille is correct. The speedo continues to show your speed and doesn’t drop to Zero.

The interim report doesn't delve into the reasons why, well have to wait for the final report.
Paragraph 83 does state that the circuit breaker which supplies current to the system was found tripped upon investigation. But it is in an area that sustained damage during the collision.
 
Sponsor Post - registered members do not see these adverts; click here to register, or click here to log in
R

RailUK Forums

43096

On Moderation
Joined
23 Nov 2015
Messages
18,722
We have a bit more info about my questions from a few months ago about the radio communications from the incident train which makes for interesting reading:



So the driver of 1J25 did use their radio to alert the signaller to the slide which allowed the signaller to phone up the driver of 1S71 to alert them which no doubt helped reduce the severity of the collision as it will have seen 1S71 slow down earlier than it might otherwise have done so with no warning. But I do still wonder why the driver of 1J25 didn't make a Railway Emergency Call instead? Feels like that would be a situation in which to hit the red button rather than phone up the signaller? That being said perhaps it wouldn't have made that much difference in the grand scheme seeing as the driver of 1S71 was reporting low adhesion, going into full service breaking immediately might have triggered it's own slide!
Agreed. I would also question why the one shot emergency sanders were not used. SPADing a loop on a single line would seem a time to use them.

Regardless of the condition of the sanding equipment, that’s two big questions about the driver’s response.
 

millemille

Member
Joined
28 Jul 2011
Messages
407
Location
Derbyshire
Agreed. I would also question why the one shot emergency sanders were not used. SPADing a loop on a single line would seem a time to use them.

Regardless of the condition of the sanding equipment, that’s two big questions about the driver’s response.
Both of those actions should/would have taken place while the driver was in a high stress/high workload situation in the cab so some latitude has to be allowed.

However, the way the train was being driven before the Block Marker was passed, brake step 1 applications for the initial slowing of the train and ERTMS intervention seconds after the first step 2 application, when it should have been "business as usual" in the cab needs some understanding.

But nonetheless, if the automatic sander had been working* the train would** still have been able to have come to stand at, or before, the block marker when ERTMS applied full service and then emergency brake irrespective of when the driver applied the brakes.


*The RAIB are not yet saying when the automatic sander defects occurred, that will hopefully come in the final report as they are still investigating the defects, but there's enough circumstantial evidence to say with confidence that the automatic sander were not working and had not been working for some time IMO


**based on the data in the RAIB interim report, when ERTMS intervened the train would have needed to have stopped at approximately 4.5%g to not pass the block marker. It actually achieved approximately 3%g (of which about 2.5%g came from the train brakes/available adhesion and about 0.5%g came from the rising gradient). 2kg/min of sand from the automatic sander would have provided sufficient conditioning of the railhead to raise the adhesion to allow an additional 1.5%g to have been achieved (RSSB research project T1107 gives data to prove this)
 
Last edited:

43096

On Moderation
Joined
23 Nov 2015
Messages
18,722
Both of those actions should/would have taken place while the driver was in a high stress/high workload situation in the cab so some latitude has to be allowed.
I get that it is a high stress situation, and can understand if the “one shot” sander was forgotten about as the regular sanders would be the default use.

However I do not understand why a driver would phone the signaller first in a scenario where they are sliding through a loop where they know there is a train on the single line beyond. First thought should be stopping the train coming the other way - that must be a REC situation. Likewise why did the signaller not initiate a REC rather than phoning the other service? I’m afraid driver and signaller have wasted precious time with their actions.
 

PG

Established Member
Joined
12 Oct 2010
Messages
3,470
Location
at the end of the high and low roads
The RAIB are not yet saying when the automatic sander defects occurred, that will hopefully come in the final report as they are still investigating the defects, but there's enough circumstantial evidence to say with confidence that the automatic sander were not working and had not been working for some time IMO
Interesting, thanks. I wonder when the last below solebar prep occurred? I'm sure the final report will enlighten us.

Based on this, the final report may well have some kind of guidance on the interval between below solebar prep's or suggest that the RSSB investigate the specifics ofsander function further.
 

millemille

Member
Joined
28 Jul 2011
Messages
407
Location
Derbyshire
Interesting, thanks. I wonder when the last below solebar prep occurred? I'm sure the final report will enlighten us.

Based on this, the final report may well have some kind of guidance on the interval between below solebar prep's or suggest that the RSSB investigate the specifics of sander function further.
The interim report states...

"34 Before entering service on the morning of 20 October, unit 158841 was prepared by a driver instructor and a trainee driver. This preparation included internal and external inspections of the train, together with further checks of the functionality of safety systems, including the train’s automatic sanders.

35 During 20 October, unit 158841 was operated on various routes to Aberystwyth, Shrewsbury, Birmingham and Chester. After the last passenger service finished at Shrewsbury, unit 158841 was planned to be stabled overnight at Crewe maintenance depot. However, due to service disruption, the train was instead driven to Chester station and stabled in a platform, arriving just before midnight.

36 Early on the morning of 21 October, a train driver prepared the train for service. However, the train was positioned adjacent to a platform. This meant that the driver did not have access to equipment on the train’s underframe, including the sander test button. Consequently, certain safety systems, including the operation of the automatic sanding system, could not be checked."
 
Last edited by a moderator:

Annetts key

Established Member
Joined
13 Feb 2021
Messages
3,786
Location
West is best
I get that it is a high stress situation, and can understand if the “one shot” sander was forgotten about as the regular sanders would be the default use.

However I do not understand why a driver would phone the signaller first in a scenario where they are sliding through a loop where they know there is a train on the single line beyond. First thought should be stopping the train coming the other way - that must be a REC situation. Likewise why did the signaller not initiate a REC rather than phoning the other service? I’m afraid driver and signaller have wasted precious time with their actions.
This is a human factors problem. I'm sure this has come up before. Although I can't remember if it was in respect of railways (drivers, guards and signallers) or aviation. Humans in a high stress situation will often not think about using functions / facilities that are normally not used, or which are infrequently used. That includes emergency equipment and systems.

In respect of signallers, some PSB have an "all signals emergency" control which puts all controlled signals (and automatic signals that have a individual replacement switch/control) in an area back to red immediately. This is far faster than dealing with each signal individually. But in practice, these emergency facilities are hardly ever used in anger. Instead the signaller normally uses the individual signal switches/controls, because that is what they are used to.

You get the same in road vehicles, do the majority of drivers know where and how to operate little used functions / facilities? I had it myself once. The traffic on the motorway came to a sudden stop so I wanted to put my hazard lights on. But at first I couldn’t find the hazard light switch...

Humans also have difficulty dealing with a failure or potential failure of certain automatic systems. The automatic sander here being an example. Are drivers trained to identify the symptoms of a failure of such a system? Have they been trained on what to do in this situation? Have they practised this or been in a simulator to experience this?

Better training and regularly using simulators to recreate unusual situations so that staff experience these situations means there is a better chance that should they need to deal with a situation in real life, they hopefully will respond quickly and effectively.
 

edwin_m

Veteran Member
Joined
21 Apr 2013
Messages
28,586
Location
Nottingham
In respect of signallers, some PSB have an "all signals emergency" control which puts all controlled signals (and automatic signals that have a individual replacement switch/control) in an area back to red immediately.
One aviation example is the Auxiliary Power Unit, which is often used on the ground but rarely needs to be turned on during a flight.

With the "Miracle on the Hudson", Sullenberger remembered to turn the APU on when both engines failed, so still had enough power to control the aircraft. In the recent Jeju Air crash it was suggested they may not have, although I don't know if that has since been established one way or the other.
 

ainsworth74

Forum Staff
Staff Member
Global Moderator
Joined
16 Nov 2009
Messages
31,067
Location
Redcar
With the "Miracle on the Hudson", Sullenberger remembered to turn the APU on when both engines failed, so still had enough power to control the aircraft. In the recent Jeju Air crash it was suggested they may not have, although I don't know if that has since been established one way or the other.
Though isn't that also a checklist/memory items issue? Activating the APU is one of the steps to take in a situation where you lose both engines (or have some other issue meaning you lose power) but it's quite a long way down the list. I'm sure I've read or heard somewhere or other that there's arguments that the checklists need changing to make starting the APU one of the first things that crew do when they suffer a dual engine failure.

But yes it's certainly quite a good example of a situation where it's easy to overlook an obvious (in hindsight) decision to improve your situation.
 

43066

Veteran Member
Joined
24 Nov 2019
Messages
12,089
Location
London
I get that it is a high stress situation, and can understand if the “one shot” sander was forgotten about as the regular sanders would be the default use.

However I do not understand why a driver would phone the signaller first in a scenario where they are sliding through a loop where they know there is a train on the single line beyond. First thought should be stopping the train coming the other way - that must be a REC situation. Likewise why did the signaller not initiate a REC rather than phoning the other service? I’m afraid driver and signaller have wasted precious time with their actions.

You’ve surely answered your own question in the first sentence of your post - high stress situation and associated human factors. A REC call is still a call to the signaller and there’s a temptation not to jump to that initially, even if with 20/20 hindsight that would have been the best thing to have done.
 

crablab

Established Member
Joined
8 Feb 2020
Messages
1,288
Location
UK
But yes it's certainly quite a good example of a situation where it's easy to overlook an obvious (in hindsight) decision to improve your situation.
...and why I wondered whether the emergency sander is something that should be activated in a similar scenario (discussed above)

Not to stretch the aviation analogy too far, but modern aircraft have high levels of automation and particularly flight envelope protection & recovery precisely for this reason - humans are falable and take longer to process information and reach a decision.

In some cases, taking an automated & decisive action is preferred (even if it's not the most optiomal), rather than waiting for humans to come to a conclusion.

TCAS is a good example. Likewise the rollout of TPWS to augment AWS, in preference to SPAD indicators.
 

Krokodil

Established Member
Joined
23 Jan 2023
Messages
6,524
Location
Wales
Though isn't that also a checklist/memory items issue? Activating the APU is one of the steps to take in a situation where you lose both engines (or have some other issue meaning you lose power) but it's quite a long way down the list. I'm sure I've read or heard somewhere or other that there's arguments that the checklists need changing to make starting the APU one of the first things that crew do when they suffer a dual engine failure.
If the film is accurate, turning on the APU was some way down the list, whereas Sullenberger did it first.
 

edwin_m

Veteran Member
Joined
21 Apr 2013
Messages
28,586
Location
Nottingham
Not to stretch the aviation analogy too far, but modern aircraft have high levels of automation and particularly flight envelope protection & recovery precisely for this reason - humans are falable and take longer to process information and reach a decision.

In some cases, taking an automated & decisive action is preferred (even if it's not the most optiomal), rather than waiting for humans to come to a conclusion.
The train already has a system to do this - the automatic sander. This triggers when it detects a demand for braking combined with the existence of wheelslip, similar to how an aircraft system will operate or be enabled or inhibited according to a combination of factors that tells the aircraft that it is warranted. The emergency sander is then a backup to the backup and purely manual - if it was automated it would have to have the same set of triggers, plus something to detect the automatic sander not dispensing sand - probably also with a time lag to prevent it working if the automatic sander stops for a short period.

If this train had not happened to be fitted with an emergency sander then the system would never have entered into this discussion. It's fitted to a small proportion of the total 158 plus 159 fleet, plus I believe also 22x units, which are an even smaller proportion of the total train fleet.
 

millemille

Member
Joined
28 Jul 2011
Messages
407
Location
Derbyshire
The train already has a system to do this - the automatic sander. This triggers when it detects a demand for braking combined with the existence of wheelslip, similar to how an aircraft system will operate or be enabled or inhibited according to a combination of factors that tells the aircraft that it is warranted. The emergency sander is then a backup to the backup and purely manual - if it was automated it would have to have the same set of triggers, plus something to detect the automatic sander not dispensing sand - probably also with a time lag to prevent it working if the automatic sander stops for a short period.

If this train had not happened to be fitted with an emergency sander then the system would never have entered into this discussion. It's fitted to a small proportion of the total 158 plus 159 fleet, plus I believe also 22x units, which are an even smaller proportion of the total train fleet.
Removed from the 22X fleets in 2017.
 

357

Established Member
Joined
12 Nov 2018
Messages
2,282
I believe also 22x units
The 22x units I drive either don't have an emergency sander or I've never been told about it and it's not obviously labeled in the cab.
 

millemille

Member
Joined
28 Jul 2011
Messages
407
Location
Derbyshire
The 22x units I drive either don't have an emergency sander or I've never been told about it and it's not obviously labeled in the cab.
To repeat, the 220 and 221 units had the twin-shot bottles removed in 2017, as a consequence of the Chester station accident in 2016, and replaced with an automatic sander.
 

MrJeeves

Established Member
Associate Staff
Senior Fares Advisor
Joined
28 Aug 2015
Messages
4,608
Location
Burgess Hill
To repeat, the 220 and 221 units had the twin-shot bottles removed in 2017, as a consequence of the Chester station accident in 2016, and replaced with an automatic sander.
The accident in question, below, was in 2013 not 2016, right?



I see on the ORR's site that Virgin Trains had all their units fitted by October 2015. CrossCountry's units were intended to be fitted by February 2017. CrossCountry only confirmed that they were fitted across all units from 1 October 2019. Whether this was a delay in reporting completion of the implementation, or whether it did actually take that long, isn't clear.

 
Last edited:

357

Established Member
Joined
12 Nov 2018
Messages
2,282
To repeat, the 220 and 221 units had the twin-shot bottles removed in 2017, as a consequence of the Chester station accident in 2016, and replaced with an automatic sander.
Apologies - didn't see your post when I made mine.

Glad to read I'm not going crazy and we do indeed only have the normal sander :)
 

millemille

Member
Joined
28 Jul 2011
Messages
407
Location
Derbyshire
The accident in question, below, was in 2013 not 2016, right?



I see on the ORR's site that Virgin Trains had all their units fitted by October 2015. CrossCountry's units were intended to be fitted by February 2017. CrossCountry only confirmed that they were fitted across all units from 1 October 2019. Whether this was a delay in reporting completion of the implementation, or whether it did actually take that long, isn't clear.


Yes, you're right '13 not '16.

The whole fleet fitment for original 58 (?) XC 220 & 221's was complete by late 2017. The Virgin fleet had been completed earlier, but was before my involvement with the fleets. The original sander fitted to the Virgin units which have since transferred to XC has been removed and they now have the same sander as the original XC units to ensure standardization across the fleet.
 

Anonymous10

Established Member
Joined
19 Dec 2019
Messages
2,629
Location
wales
Unsure if it relates to this incident however a tfw 158 is at arriva train care in Bristol.
 

ainsworth74

Forum Staff
Staff Member
Global Moderator
Joined
16 Nov 2009
Messages
31,067
Location
Redcar
The RAIB have now released their report into this accident:
Summary
At around 19:26 on Monday 21 October 2024, train reporting number 1J25, the 18:31 service from Shrewsbury to Aberystwyth, collided with train 1S71, the 19:09 Machynlleth to Shrewsbury service. The collision took place near Talerddig, Powys. Both train services were operated by Transport for Wales Rail Limited (TfWRL).

The trains were operating on Network Rail’s Cambrian lines. This is predominantly single track, with passing loops provided at intervals to allow trains travelling in opposite directions to pass each other. Train 1J25 had been approaching the loop at Talerddig, with an intended stopping position within the loop, so that train 1S71 could pass it. However, train 1J25 did not stop as intended, continuing through the passing loop and into the single line section on which train 1S71 was approaching.

The head-on collision occurred while train 1J25 was travelling at approximately 39 km/h (24 mph) and train 1S71 was travelling at approximately 11 km/h (6 mph) in the opposite direction. At the point where the collision occurred, train 1J25 had travelled around 1,080 metres beyond its intended stopping position.

A passenger travelling on train 1J25 suffered fatal injuries as a consequence of the collision. Three people on this train were also seriously injured, including the train’s guard, and 18 other people received minor injuries. The driver of train 1S71 suffered serious injuries, and the remaining five people on this train all suffered minor injuries.

Neither train derailed, but extensive damage was caused to both trains. Damage was also caused to railway infrastructure in Talerddig loop. The railway remained closed until 28 October 2024.

RAIB’s investigation found that the accident was caused by train 1J25 passing its authorised stopping position and entering the single line beyond, which was occupied by the approaching train 1S71.

Train 1J25 passed its authorised stopping position due to a combination of three factors. These were that the wheel-rail adhesion in the area approaching the Talerddig loop was low, although not exceptionally so for this area during October. In addition, the two sanding systems fitted to train 1J25 which could have mitigated the prevailing low adhesion conditions, and avoided the accident, did not dispense sand. The automatic sander did not function, probably due to the presence of electrical faults in its control circuit, while the manually operated emergency sander was not activated by the driver. The third factor was that the approach speed of train 1J25 towards the eastern entry to Talerddig loop was such that the deceleration required to slow the train for the loop could not be sustained with the available wheel-rail adhesion.

Having passed its intended stopping position within Talerddig loop, train 1J25 entered the single line beyond. This area had exceptionally low wheel-rail adhesion and was on a steep downhill gradient. This meant that, even though the brakes on train 1J25 remained applied, the train did not decelerate as it approached train 1S71. There were no engineered mitigations to prevent train 1J25 entering the occupied single line in the event of an overrun.

Behind this lay a number of underlying factors and factors affecting the consequences of the accident. These involve a range of issues and organisations.

Recommendations
As a result of the investigation, RAIB has made nine recommendations:

Two recommendations are made to the Rail Safety and Standards Board and Angel Trains to improve the design, maintenance and testing of trainborne sanding equipment.
A recommendation is made to Network Rail to review the assumptions made to justify the use of simple assessments of overrun risk on the Cambrian lines, considering the circumstances of this accident, and updated industry standards.
A recommendation is made to Network Rail to improve overrun protection in future versions of software-based train control systems and for it to review how overrun risks are assessed, considering the circumstances of this accident.
A further recommendation is made to Network Rail to improve wheel-rail adhesion conditions through the application of an improved understanding of the effectiveness of railhead treatment regimes. This builds on a previous RAIB recommendation following the 2021 accident at Salisbury.
Two recommendations are made to TfWRL to ask it to review how drivers are trained, based on issues identified in this investigation.
A recommendation is made to the Rail Safety and Standards Board to review standards and rules governing the design of passenger train interior fittings to reduce the risk to passengers in the event of an accident.
A recommendation made to TfWRL intends that all on-train staff, irrespective of role have the skills and knowledge required to assist in the event of an emergency.
RAIB also identified a learning point relating to reaching a clear understanding when safety‑critical communications take place between signallers and train drivers.

 

Purple Train

Established Member
Joined
16 Jul 2022
Messages
2,444
Location
Despond
My first reaction on reading the report is that, as with the recent near miss with track workers at Bookham that was the subject of the last RAIB report (discussion here), there has been an absolutely huge amount of Swiss cheese aligning here to create this dreadful incident.

The report seems to be unusually explicit that the error of the driver of 1J25 in not operating the emergency sander was a factor in the incident, but equally they do, in a more typical and subtle RAIB way, hint that the driver was set up to fail to an extent.

I think it's telling that the RAIB were explicit about the sander, but somewhat glossed over the fact that 1J25 was being controlled in a way typical of a train passing along the main line rather than through the loop, because of the confusing way in which speed limits were presented on the ERTMS, and the driving style which had been adopted on the line. Compare, for instance, the tone of paragraph 239 of the report, and the tone of paragraph 204. In particular, I found it surprising that drivers reported getting interventions from the system at least once a week (paragraph 368). That suggests to me that this was an accident waiting to happen. Also, if that kind of culture existed on the line with regard to driving style, with a huge amount of trust placed on the ERTMS system, I can completely understand how the unfortunate driver of 1J25 forgot about a manual system which they never had cause to use and didn't seem to have been adequately trained on anyway, particularly given they had no reason to believe that the automatic system wasn't working. I suppose, though, that there is no obvious solution to the fact that the Cambrian is worked entirely by drivers who only sign about three minutes'-worth of conventionally-signalled track, given the length of the line. But it can't have helped matters that the Cambrian is so thoroughly separate to any other railway environment.

I am slightly disappointed that the RAIB gave mention to seatbelts in paragraph 329 without explicitly saying that if the poor driver of 1S71 had been wearing a seatbelt (as was mooted after Carmont) the consequences may well have been even worse for them given the state of their cab.

A final word on the severity of the collision; given the excellent safety record of railways in the UK I think it's a stark reminder that a 30mph head-on collision carries a huge amount of energy, and thankfully we haven't had such an incident in years. Hopefully, the various small lessons learned from this incident, which combined in such a terrible way, won't need to be learnt in blood again.

RAIB report references:
Para 239
Although this comparison uses a small sample of drivers, it indicates that train 1J25 was being driven on the day of the accident in a manner similar to that adopted by other drivers when following the higher speed profile through the Down Loop.
Para 204
The driver of train 1J25 passed two block markers applying to westbound trains during the overrun. The first was MH1078, located in the loop, which was the end of the movement authority for the train. The second was block marker MH1080, located 259 metres beyond the start of the single line (figure 4). To comply with the Rule Book and TfWRL braking instructions, the driver of train 1J25 should have discharged the emergency sanders on realising the train would pass block marker MH1078.
Para 368
During the investigation, RAIB found that TfWRL was not aware of how frequently SBIs were occurring. However, RAIB was provided with witness evidence from drivers that they could experience an SBI as frequently as once a week. This is supported by data taken from the JRU on board unit 158824 which indicates that 17 SBIs had occurred in the 21 days preceding the accident on that unit alone.
Para 329
Class 158 vehicles, in common with all other passenger rail vehicles in the UK and in other countries, are not fitted with seat belts. This means that in end-on collisions, train occupants are vulnerable to secondary impact injuries as a result of being thrown around the vehicle interior. Except for the driver of train 1S71, all the injuries sustained during the accident were assessed as being caused by secondary impact. Seat belts could have been effective in reducing the severity of the secondary impact injuries if they had been fitted and worn by the occupants at the time of the accident.
 

Dave W

Member
Joined
27 Sep 2019
Messages
874
Location
North London
I concur with the excellent summary above - as always what they don't say is almost more telling than what they do.

The analysis of the warning process of overspeed pre-intervention (paras 211-229) was very interesting, and the number of contributory factors is very high, especially when, on the face of it, the trains collided because one simply could not stop. Such is the complexity of these incidents, I suppose.
 

sefyllian

Member
Joined
14 Jan 2013
Messages
138
Now I have learned how Talerddig should be pronounced. Stress on the middle syllable, like "allergic", not the last as in Tallyllyn or Ynys Mon
Correct! Welsh words of more than one syllable are always (with very few exceptions) stressed on the penultimate syllable. Tal-y-llyn is more properly written like so (in Welsh at least), for that reason (ie. to show it’s stressed as three one syllable words, rather than as one three syllable word, which is why it’s different to Talerddig)
 

norbitonflyer

Established Member
Joined
24 Mar 2020
Messages
5,657
Location
SW London
Correct! Welsh words of more than one syllable are always (with very few exceptions) stressed on the penultimate syllable. Tal-y-llyn is more properly written like so (in Welsh at least), for that reason (ie. to show it’s stressed as three one syllable words, rather than as one three syllable word, which is why it’s different to Talerddig)
So is the Other Railway really Ffesti-NEE-og?, or is the "io" treated as one syllable (Ffes-TIN-yog) ?
 

Top