Report of collision Bedford (Elstow) - 19/06/26

hexagon789 · 8 Oct 2026

WAO said:
Didn't note my source for the speed but as the SA says 50, it's 50.
That still suggests a junction indication on the slow of double yellow or flashing yellow, rather than a true distant (for a red).
WAO

It's approach release from red.

Therefore, you can't receive a clear or even caution indication initially if diverging even if the line ahead is clear.

There's no 'flashers' fitted either; if memory serves the speed differential wouldn't meet the standards for flashing yellows anyway.

So, the train approaches at caution with a single yellow at the platform signal.

Assuming the crossing move can be set and at least first block on the Up Fast is clear then the junction signal steps up to a single yellow once the train has occupied the track circuit approaching the junction signal for long enough - I.e. deemed approaching sufficiently slowed.

After reaching the AWS magnet for the signal, for which a 'caution' will be relieved, then the signal will then step up to the appropriate aspect for the line ahead including clear.

Therefore, the most common signal sequence on approach if taking the divergence from Up Slow over to the Up Fast would be:

- train approaches junction signal at danger
- signal steps up to yellow on approach
- train reaches magnet and receives AWS caution
- signal steps up to green

The actual sequence would seem to be that the junction signal remained at single yellow, as the signal after the crossover onto the Up Fast was at danger still.

RailUK Forums

Teapot42 · 8 Oct 2026

fishwomp said:
flash isn't new - the 810 is new, the 360 isn't as new, but no reason it shouldn't be upgraded at mid-life etc.
...
A 256GB SD card HD dashcam would store over 24 hours, no problem. As said by another, it's the last few mins that matter - although reviewing footage over previous days was useful in the recent heatwave landslips..

Flash has much lower write endurance than traditional hard drives. CCTV involves storing a *lot* of data, especially on trains where it's likely to be running much of the time, certainly considerably in excess of a dashcam. It may well be a technical review of the available technologies highlighted that flash memory with sufficient endurance wasn't available at design time.

It's worth noting that magnetic tape is still widely used for backing up data, so just because a particular technology has replaced something in one application doesn't mean it's going to in all applications.

It's quite common to buffer data and write in chunks, not least because a power cut during a write operation can cause significant corruption. That said, it wouldn't be too surprising if a design assumption was that power would always be present, or that the train / system would usually be shut down in an orderly manner to avoid the risk of corruption. Missing as much as two minutes seems unusual to me, but without knowing the design spec / assumptions, and whether that data wasn't stored or couldn't be recovered it's hard to know what to make of it.

Unobrow · 8 Oct 2026

Efini92 said:
Harrow is different though as there was fog.

Quintinshill?

35B · 8 Oct 2026

Teapot42 said:
Flash has much lower write endurance than traditional hard drives. CCTV involves storing a *lot* of data, especially on trains where it's likely to be running much of the time, certainly considerably in excess of a dashcam. It may well be a technical review of the available technologies highlighted that flash memory with sufficient endurance wasn't available at design time.

It's worth noting that magnetic tape is still widely used for backing up data, so just because a particular technology has replaced something in one application doesn't mean it's going to in all applications.

It's quite common to buffer data and write in chunks, not least because a power cut during a write operation can cause significant corruption. That said, it wouldn't be too surprising if a design assumption was that power would always be present, or that the train / system would usually be shut down in an orderly manner to avoid the risk of corruption. Missing as much as two minutes seems unusual to me, but without knowing the design spec / assumptions, and whether that data wasn't stored or couldn't be recovered it's hard to know what to make of it.

I worry about the quality of the design process if that was the assumption. It is absolutely not fail safe and, worse, is designed not to record some of the most significant data.

20+ years ago, a cousin had a helmet mounted camera that was always on, with data overwritten every couple of minutes. That was consumer grade equipment.

norbitonflyer · 8 Oct 2026

Unobrow said:
Quintinshill?

Not a SPAD. The two moving trains were both running under clear signals, but the up (southbound) troop train was on a collision course with a down stopping train that had been shunted onto the up line to let a down express overtake it. Wreckage from the initial collision on the up line, between the up troop train and the down local, obstructed the down line just as the down express arrived on the scene. Two other trains standing in the up and down goods loops were also damaged.

The failure of the signalmen (plural, there was a shift change at the crucial time) to remember the local was there, despite one of them having signalled the manouvre, and his relief having actually arrived on board it, was the cause of the accident

Vexed · 8 Oct 2026

35B said:
I worry about the quality of the design process if that was the assumption. It is absolutely not fail safe and, worse, is designed not to record some of the most significant data.

20+ years ago, a cousin had a helmet mounted camera that was always on, with data overwritten every couple of minutes. That was consumer grade equipment.

You can get flash storage that is rated for it to be written to >1000x it's capacity. I have a few Micron 5400 Pros with a storage size of 960GB and they're rated for 2628 TBW (terabytes written) which means the whole drive can be rewritten 1.5 times a day for the whole 5 year warranty period. The Max version of that drive is rated for 5 full writes per day for 5 years.

CCTV drives (unless backed up elsewhere but that's a lot of bandwidth) would probably be written to in full once a month if you need retention for 30 days so unless they expect the flash to last for the whole ~40 years of a trains life I struggle to see it being a problem.

Flash technology has improved massively over the last 15 years, so I understand why it wasn't viable in the past. Good quality enterprise drives last ages these days.

Unobrow · 8 Oct 2026

norbitonflyer said:
Not a SPAD. The two moving trains were both running under clear signals, but the up (southbound) troop train was on a collision course with a down stopping train that had been shunted onto the up line to let a down express overtake it. Wreckage from the initial collision on the up line, between the up troop train and the down local, obstructed the down line just as the down express arrived on the scene. Two other trains standing in the up and down goods loops were also damaged.

The failure of the signalmen (plural, there was a shift change at the crucial time) to remember the local was there, despite one of them having signalled the manouvre, and his relief having actually arrived on board it, was the cause of the accident

Of course, you’re totally correct. I blame tonight’s red wine for my foggy brain!

Teapot42 · 9 Oct 2026

Vexed said:
You can get flash storage that is rated for it to be written to >1000x it's capacity. I have a few Micron 5400 Pros with a storage size of 960GB and they're rated for 2628 TBW (terabytes written) which means the whole drive can be rewritten 1.5 times a day for the whole 5 year warranty period. The Max version of that drive is rated for 5 full writes per day for 5 years.

CCTV drives (unless backed up elsewhere but that's a lot of bandwidth) would probably be written to in full once a month if you need retention for 30 days so unless they expect the flash to last for the whole ~40 years of a trains life I struggle to see it being a problem.

Flash technology has improved massively over the last 15 years, so I understand why it wasn't viable in the past. Good quality enterprise drives last ages these days.

It's understandably common to misunderstand the difference between commercial and industrial applications, and how technology is implemented between them.

Industrial applications are rated to wider temperature ranges, for longer lifetimes, and often with availability guarantees spanning a decade or more. Those factors limit the ability to use consumer-grade parts. These are often rated across a lower temperature range (say 0-70 rather than -40 to +125), and there is no guarantee you'll be able to buy one tomorrow, as the factory will stop producing them once demand dwindles.

I'd imagine there will be some more flexibility to change drive used on a CCTV system than the industrial systems I'm used to designing, but even then it's likely the manufacturer will want to do their own performance testing on whatever drive is used, rather than just picking one off the shelf at random. Therefore, they are more likely to talk to the drive manufacturers and get guarantees of production life for particular units.

This then ties back to the CCTV system used - was it designed specifically for the 810, in which case it could well be using late 2010s technology. Even if it's an off-the-shelf unit, the design is likely semi-frozen several years in the past.

Either way, it's possible the certification it holds is based on using certain specific components. If it's supplied as a complete system with drive installed then tests have to be performed on the finished unit. Changing drive means those certifications need repeating, which is costly.

All these factors mean that it's quite likely the flash memory used is several generations old, and is designed to last the service life of the train, which will be several decades.

The fact a consumer-grade drive is available now which may meet those requirements doesn't mean an industrial grade drive from 5 years ago can.

WAO · 9 Oct 2026

So no ambiguity.
WAO

Vexed · 9 Oct 2026

Teapot42 said:
It's understandably common to misunderstand the difference between commercial and industrial applications, and how technology is implemented between them.

Industrial applications are rated to wider temperature ranges, for longer lifetimes, and often with availability guarantees spanning a decade or more. Those factors limit the ability to use consumer-grade parts. These are often rated across a lower temperature range (say 0-70 rather than -40 to +125), and there is no guarantee you'll be able to buy one tomorrow, as the factory will stop producing them once demand dwindles.

I'd imagine there will be some more flexibility to change drive used on a CCTV system than the industrial systems I'm used to designing, but even then it's likely the manufacturer will want to do their own performance testing on whatever drive is used, rather than just picking one off the shelf at random. Therefore, they are more likely to talk to the drive manufacturers and get guarantees of production life for particular units.

This then ties back to the CCTV system used - was it designed specifically for the 810, in which case it could well be using late 2010s technology. Even if it's an off-the-shelf unit, the design is likely semi-frozen several years in the past.

Either way, it's possible the certification it holds is based on using certain specific components. If it's supplied as a complete system with drive installed then tests have to be performed on the finished unit. Changing drive means those certifications need repeating, which is costly.

All these factors mean that it's quite likely the flash memory used is several generations old, and is designed to last the service life of the train, which will be several decades.

The fact a consumer-grade drive is available now which may meet those requirements doesn't mean an industrial grade drive from 5 years ago can.

The Micron 5400 I used in my example isn't consumer grade, it's an enterprise grade drive, though it's not industrial grade so not rated for massive temperature swings or heavy vibration.

I wouldn't be surprised if the Class 810 CCTV has some design similarities to the 800s, so could be a design from over 10 years ago.

35B · 9 Oct 2026

Teapot42 said:
It's understandably common to misunderstand the difference between commercial and industrial applications, and how technology is implemented between them.

Industrial applications are rated to wider temperature ranges, for longer lifetimes, and often with availability guarantees spanning a decade or more. Those factors limit the ability to use consumer-grade parts. These are often rated across a lower temperature range (say 0-70 rather than -40 to +125), and there is no guarantee you'll be able to buy one tomorrow, as the factory will stop producing them once demand dwindles.

I'd imagine there will be some more flexibility to change drive used on a CCTV system than the industrial systems I'm used to designing, but even then it's likely the manufacturer will want to do their own performance testing on whatever drive is used, rather than just picking one off the shelf at random. Therefore, they are more likely to talk to the drive manufacturers and get guarantees of production life for particular units.

This then ties back to the CCTV system used - was it designed specifically for the 810, in which case it could well be using late 2010s technology. Even if it's an off-the-shelf unit, the design is likely semi-frozen several years in the past.

Either way, it's possible the certification it holds is based on using certain specific components. If it's supplied as a complete system with drive installed then tests have to be performed on the finished unit. Changing drive means those certifications need repeating, which is costly.

All these factors mean that it's quite likely the flash memory used is several generations old, and is designed to last the service life of the train, which will be several decades.

The fact a consumer-grade drive is available now which may meet those requirements doesn't mean an industrial grade drive from 5 years ago can.

All of the above may be true, but the reality of these decisions is that an important investigative tool is not available for reviewing the critical period before an incident. That, to me, is a fundamental flaw in the requirements.

fishwomp · 9 Oct 2026

35B said:
All of the above may be true, but the reality of these decisions is that an important investigative tool is not available for reviewing the critical period before an incident. That, to me, is a fundamental flaw in the requirements.

This.

I think we've done flash storage to death now! It's a solved problem. Alas the requirements are not strong enough to actually require the solution to the need.

Presumably it is not a mandatory piece of kit - except on new stock - would a steam loco have it?

The other thing missing is cab CCTV - I haven't heard anything said of that. We're in the dark.

Ah:

TRAIN DRIVER IN CAB ATTENTION & ALERTNESS MONITORING DEVICE

www.rmt.org.uk

dated August 2025.

... RSSB trial, “Driver Attention and Alertness IMP-T1193”.

The trial is running on LNER, C2C and DB Cargo operated trains. The equipment (provided by a company called ‘Seeing Machine’) and is installed on a limited number of trains and activates when the driver shows signs of fatigue.

but here we go:

I have notified RSSB T1193 project leads that RMT oppose the use of cameras in driving cabs – and that I am concerned as to the creeping introduction of digital monitoring and surveillance cameras relating to in cab technology.

RMT say ASLEF also has issues - but I couldn't find their policy on it.

One would think that in-cab facing CCTV that could only be opened by BTP for evidence in a specific involved vehicle would be a solution to the monitoring fears.

Yet what about platform staff... they are watched every minute of their shift by cameras. (edit: so are train managers as they walk through the passenger cars)

The Air India accident didn't have cockpit video - only sound, so it's not just a railways problem.

3RDGEN · 9 Oct 2026

35B said:
All of the above may be true, but the reality of these decisions is that an important investigative tool is not available for reviewing the critical period before an incident. That, to me, is a fundamental flaw in the requirements.

The CCTV data is "missing" for the last 3 or 4 seconds but we know that 1H46 had already pass the protecting signal at red and the brakes had been applied so I can't see what we are missing? Unless you plan to replace CCTV on the entire national fleet every few years you will always have old technology, given design/build times are 3-4 years it's likely the kit you fit is obsolete at delivery of a new fleet anyway.

fishwomp said:
The other thing missing is cab CCTV - I haven't heard anything said of that. We're in the dark.

The unions will not accept it so it's not worth going there.

35B · 9 Oct 2026

3RDGEN said:
The CCTV data is "missing" for the last 3 or 4 seconds but we know that 1H46 had already pass the protecting signal at red and the brakes had been applied so I can't see what we are missing? Unless you plan to replace CCTV on the entire national fleet every few years you will always have old technology, given design/build times are 3-4 years it's likely the kit you fit is obsolete at delivery of a new fleet anyway.

The unions will not accept it so it's not worth going there.

If I recall, it's not 3-4 seconds, but a couple of minutes that are missing.

I find the union view interesting and depressing in what it says about industrial relations. More generally, there's a strong part of me that questions the role of film where it is most often going to be required in response to fatalities that it captures. In a similar way that I'm revolted by news from the Pentagon overnight, I'm not convinced that we as a society have a right to compel people to have to watch other people in life threatening incidents.

AlterEgo · 9 Oct 2026

35B said:
If I recall, it's not 3-4 seconds, but a couple of minutes that are missing.

I find the union view interesting and depressing in what it says about industrial relations. More generally, there's a strong part of me that questions the role of film where it is most often going to be required in response to fatalities that it captures. In a similar way that I'm revolted by news from the Pentagon overnight, I'm not convinced that we as a society have a right to compel people to have to watch other people in life threatening incidents.

FFCCTV and in cab CCTV are, and will be, used for many other reasons than for life threatening incidents. Nobody will be *compelled* to watch it, but I would suggest if you are a RAIB investigator and you have an aversion to watching a video of someone being hit by a train you might be in the wrong job. Train crashes are not for the squeamish.

Pakenhamtrain · 9 Oct 2026

fishwomp said:
One would think that in-cab facing CCTV that could only be opened by BTP for evidence in a specific involved vehicle would be a solution to the monitoring fears.

I know on HCMT's and XT2's here the in cab cameras are only to be accessed for investigation purposes only. Some mananger can't come along and pull it to have a look

fishwomp · 9 Oct 2026

35B said:
I find the union view interesting and depressing in what it says about industrial relations. More generally, there's a strong part of me that questions the role of film where it is most often going to be required in response to fatalities that it captures. In a similar way that I'm revolted by news from the Pentagon overnight, I'm not convinced that we as a society have a right to compel people to have to watch other people in life threatening incidents.

Pathologists and police have to see things, medics too, that most people would never wish to see - they know that that is something that their duty calls upon them to do.

Rail drivers also get no say in whether someone will - intentionally or otherwise - die in front of their train.

I've never seen an impact video from a front-facing railway CCTV camera in a suicide or level crossing fatality - but you can be sure that this is something the police and investigators have to do. This is a vital piece of evidence - and comes with the duty to know what happened and, if possible, prevent something in future.

Teapot42 · 9 Oct 2026

Vexed said:
The Micron 5400 I used in my example isn't consumer grade, it's an enterprise grade drive, though it's not industrial grade so not rated for massive temperature swings or heavy vibration.

I use the term consumer broadly here. In essence, the drives you are talking about can be swapped out if necessary, that's not normally the case in industrial systems.

== Doublepost prevention - post automatically merged: 9 Oct 2026 ==

35B said:
All of the above may be true, but the reality of these decisions is that an important investigative tool is not available for reviewing the critical period before an incident. That, to me, is a fundamental flaw in the requirements.

You frame it as a poor design decision. I'm trying to point out it may be unavoidable due to the technology available. Or it could be the case there were compromises that needed to be made and someone determined that loss of power was less likely to happen than some alternative. Maybe there was a determination that frequent writes would raise the risk of corruption beyond an acceptable level.

Without being in the design team it is impossible to know, but I'm sure all these factors will have been weighed up and the best outcome in the circumstances delivered.

In this case the compromises have weighed against the desirable outcomes. That's not to say there haven't been or won't be cases where the compromises have been proved justified.

saismee · 9 Oct 2026

fishwomp said:
I've never seen an impact video from a front-facing railway CCTV camera in a suicide or level crossing fatality - but you can be sure that this is something the police and investigators have to do. This is a vital piece of evidence - and comes with the duty to know what happened and, if possible, prevent something in future.

Someone has to decide whether it is deemed suspicious or not.

3RDGEN · 9 Oct 2026

35B said:
If I recall, it's not 3-4 seconds, but a couple of minutes that are missing.

I find the union view interesting and depressing in what it says about industrial relations. More generally, there's a strong part of me that questions the role of film where it is most often going to be required in response to fatalities that it captures. In a similar way that I'm revolted by news from the Pentagon overnight, I'm not convinced that we as a society have a right to compel people to have to watch other people in life threatening incidents.

I can't find any reference to minutes only a few seconds, the FFCCTV on the CL360 was recorded to just prior to collision and provides all the information required. The 810 had a saloon CCTV fault in one vehicle much earlier in the day but that is of no relevance to the accident.

JamesT · 9 Oct 2026

3RDGEN said:
I can't find any reference to minutes only a few seconds, the FFCCTV on the CL360 was recorded to just prior to collision and provides all the information required. The 810 had a saloon CCTV fault in one vehicle much earlier in the day but that is of no relevance to the accident.

I believe the poster is referring to the rear-facing CCTV on the 810. From the report:

54 Train 1B67 first becomes visible in FFCCTV images from train 1H46 at 17:13:13.
This footage ends approximately 2 seconds later. The available rear-facing
CCTV from train 1B67 ends before this train came to a stand, so images of the
approaching train 1H46 are not captured.

The 810 came to a stand at 17:11:31 (para 48), impact was at 17:13:35 (para 57).

bramling · 9 Oct 2026

JamesT said:
I believe the poster is referring to the rear-facing CCTV on the 810. From the report:

The 810 came to a stand at 17:11:31 (para 48), impact was at 17:13:35 (para 57).

Can’t help but feel some of this is a bit academic. Whilst there’s a remote chance the rear camera on the 810 might have shed some light on what was happening inside the 360 cab, the main period of interest was whilst the 360 was between Bedford station and the red aspect that it disregarded.

Hairyspotter · 9 Oct 2026

I don't think many train drivers have an objection to in-cab cameras were they to be solely for the purposes of accident investigation. However, on TOC and FOCs that did trial them, they soon became used for purposes other than that, and therein lies the objections as I understand it

3RDGEN · 9 Oct 2026

Hairyspotter said:
I don't think many train drivers have an objection to in-cab cameras were they to be solely for the purposes of accident investigation. However, on TOC and FOCs that did trial them, they soon became used for purposes other than that, and therein lies the objections as I understand it

That's the issue, hard learnt lessons from OTMR/OTDR fitment which was sold as one thing and became the spy in the cab.

Belperpete · 9 Oct 2026

35B said:
I find the union view interesting and depressing in what it says about industrial relations. More generally, there's a strong part of me that questions the role of film where it is most often going to be required in response to fatalities that it captures. In a similar way that I'm revolted by news from the Pentagon overnight, I'm not convinced that we as a society have a right to compel people to have to watch other people in life threatening incidents.

If you read the RMT statement in full, I can understand their concerns. What was being trialled was NOT something that would only be used in the event of an incident, it was actively monitoring drivers performance. They seem to have regarded it as the thin end of the wedge.

AndrewE · 9 Oct 2026

Belperpete said:
If you read the RMT statement in full, I can understand their concerns. What was being trialled was NOT something that would only be used in the event of an incident, it was actively monitoring drivers performance. They seem to have regarded it as the thin end of the wedge.

(Apart from the fact that RMT don't represent most drivers on BR - LU is probably different) I would say they they were doing what they are paid for! Fair enough, in my mind.

Starmill · 9 Oct 2026

AndrewE said:
(Apart from the fact that RMT don't represent most drivers on BR - LU is probably different) I would say they they were doing what they are paid for! Fair enough, in my mind.

The in-cab CCTV discussion is well-worn by now and has been done to death elsewhere. It seems unlikely that much can be gained in understanding this incident from re-hashing it on this thread.

66701GBRF · 9 Oct 2026

Hairyspotter said:
I don't think many train drivers have an objection to in-cab cameras were they to be solely for the purposes of accident investigation. However, on TOC and FOCs that did trial them, they soon became used for purposes other than that, and therein lies the objections as I understand it

Exactly right. This is the issue we have recently had with groundstaff issued body cameras. They are supposed to be viewed only in the event of an incident and with restricted access. However there have been occasions where the company have gone back further than the initial incident and they even wanted to change the scope so body cameras can be used for general assessments and monitoring. Thankfully ASLEF put a stop to that.

Mag_seven · 9 Oct 2026

OK as we are now starting to speculate again, we shall close this thread and wait until the RAIB final report is issued.

thanks everybody.

Report of collision Bedford (Elstow) - 19/06/26

Veteran Member

RailUK Forums

Member

Member

Established Member

Established Member

Member

Member

Member

Established Member

Member

Established Member

Established Member

Member

Established Member

Verified Rep - Wingin' It! Paul Lucas

Established Member

Established Member

Member

Established Member

Member

Established Member

Veteran Member

Member

Member

Established Member

Established Member

Veteran Member

Member

Forum Staff