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

[baz962]

I cannot recall the document, or date, when it was, I think, effectively banned. I did manage to implement something similar a few years ago for very specific sighting reasons, but it was permitted only on the basis that the aspect would be unrestricted before the train reached the AWS, so the AWS did not 'lie' to the driver.

I've managed to check some old drawings for the Bedford area, and they confirm that 356 does (or at least did originally, it may have been changed since of course) have a two stage aspect release. Release from R to Y took place some distance from the signal, 'better aspects' require the immediate berth TC occupied, which happens after passing the AWS.

So, under these conditions, drivers will be used to getting Y + Pos 4 initially, then the aspect may (and probably most often will) step up as they get nearer but they will be cancelling the AWS warning every time, irrespective of what aspect they end up with.

I will be very surprised if that goes uncommented upon in the final report, assuming it still works that way.

@Signal Head . Sent you a pm mate.

 

[Bald Rick]

Well, ultimately, most of the past decisions on ATP, TPWS and ETCS/ERTMS come down to cost, and hence it's a political decision.

Strictly speaking, it’s down to value.

 

[bramling]

I agree, and I think I would notice that deceleration. I

All the talk of signalling systems and AWS is fascinating to a signal layperson like myself, so thanks everyone for that.
But as a non train driver, what I'm struggling with is trying to imagine a set of circumstances where a speed of 76mph when the aspect of the previous 4 aspect signal was single amber is explicable. Can anyone help me with that? Can it only be as a result of presumption?

We don’t yet have enough information to be able to fully answer that.

A more logical sequence of events is that the driver responded to the single yellow at the previous signal, correctly controlled the train over the 50mph crossovers, and then for whatever reason forgot that they were still running on a single yellow (or even hadn’t taken in that they were on a single yellow at all), and “opened up” as if on clear signals, like would be quite common at this location. However one would then expect the red signal to be noticed, and a some kind of brake application to have followed, which would have resulted in a reportable SPAD, but probably not a collision (though could have done if the 810 was stopped further north than it apparently was).

However, as you say, we have the complication that the train has driven through WH154, yet the driver does seem to have then been alert enough to have reacted pretty promptly to seeing the train standing ahead, which given it was on a curve suggests a pretty decent level of alertness.

We may never get a definite answer to this, in the same way we didn’t at Ladbroke Grove, and didn’t really at Cowden either, although in both cases there were some ideas floated as being considered highly likely (sunlight and thick fog respectively), and in the case of Cowden there was some debate as to how well lit the red aspect was and even more debate as to whether the train’s AWS was working.

For now, a lot depends on whether the driver at Bedford slowed down as if to stop and then re-accelerated past WH154, or whether he accelerated immediately after the crossovers and didn’t make any response to WH154 at all, or possibly the driving was erratic. This may give a clue as to what was going on, but still may not answer the question as to why. Unless something definitive comes to light, RAIB may have difficulty answering this one.

 

[Banana]

• Platform signal at Yellow
• Junction signal at Yellow

Why would the platform signal been at yellow?

 

[dk1]

Why would the platform signal been at yellow?

Isn’t the junction signal approach controlled? Many are.

 

[John R]

It is also from memory the first major crash involving two modern units at any great speed. The analysis of how they performed will also be an interesting part of the investigation.

There appear to be two separate strands of this investigation: what caused it to happen, and how did the units fare and what lessons can be learnt in terms of future rolling stock design and interiors.

I could imagine the second might take longer than the first, and it would be unfortunate if the second aspect considerably delayed the final report. Maybe there is scope to issue two reports?

 

[norbitonflyer]

Under discussed but mentioned above is where 5Z14 was during all this, coming the other way.

The timings put it quite possibly alongside the 810 at around the time the 810 would have become visible, and there's a possibility it could have contributed to visibility.

If it was on the down fast, it could have hidden the 810.

 

[baz962]

Isn’t the junction signal approach controlled? Many are.

Is the correct answer.

 

[Bald Rick]

Why would the platform signal been at yellow?

Because WH356 was at Red. And it was at Red because a route could not be set to WH154 on the Up Fast, as the first train was still passing / occupying the section ahead up to WH154.

 

[dk1]

Is the correct answer.

Thank you. Was quite obvious to me.

 

[hexagon789]

Why would the platform signal been at yellow?

Because the junction signal is held at red until the train is deemed sufficiently slowed to pass safely over the junction (known as approach release from red), in practice if the route is clear over the junction it will step up to single yellow while the train is approaching at perhaps 20mph or just under.

As the train nears the junction signal, if the route is clear, then when the train passes over the AWS magnet, the junction signal steps up to the suitable aspect, even green.

This latter aspect is the now no longer applied two-stage release, I.e. it exists only on older installations - it isn't current practice.

In this case, the junction signal was a read anyway, due to the presence of the 810 on the Up Fast, once that had cleared the overlap, then the route to the Up Fast could be cleared for the 360.

 

[Siemens Staines]

Under discussed but mentioned above is where 5Z14 was during all this, coming the other way.

The timings put it quite possibly alongside the 810 at around the time the 810 would have become visible, and there's a possibility it could have contributed to visibility.

Indeed. It looks like 5Z14 was checked, losing 3 minutes, on the approach to WH343 to allow 1H46 to cross in front of it. I am sure that the CCTV from this unit will form part of the RAIB investigation as it must have seen both trains.

 

[Mile Post 13]

That suggests a 'two stage' approach release, which I am aware has been done in the past but is generally out of favour now. I wouldn't have expected it to have survived as a principle long enough to have been applied on the West Hampstead scheme, but it's possible I suppose.

I was involved in the original commissioning of West Hampstead PSB.. At the time it was the first resignalling scheme to apply the then new provisions of Standard Signalling Principle 6. This permitted the signal to step up to the to the least restrictive aspect once the train had passed the AWS magnet. In the case of what was then known as Kempston Road Jct, which had 50 mph crossovers, the protecting signal displayed a single yellow with a position 4 jct indicator, releasing upwards after the train had passed the AWS magnet depending on the aspect in WH154.

Previous arrangements such as on WCML at junctions was to release from Red by means of a bi-metlalic relay which operated when the approach track circuit was occupied. On higher speed junctions such as Ledburn if trains were following one another in close proximity, the bi-metallic relay would not cool down sufficiently and so a release could (and I witnessed this) be released almost immediately on track circuit occupation. Berth track circuits on the older PSB were only provided for signals with swinging or restricted overlaps. A good example was Denbigh Hall South Jct where an overlap could be set towards the flyover and the route then called or the overlap swung back along the through route with the train approaching, however when the berth track was then occupied the overlap was locked up for three minutes.

 

[dk1]

Another signal sequence many don’t always understand is that it can be for say G, YY, YY, Y, Y, R if signal sections are short. The up & down electric lines between Stratford & Shenfield are a prime example of this.

 

[bramling]

Not railway but we had something similar in critical safety systems. To describe similar in railway context of this accident. A) something causes Aws fault and front train comes to stop. B) Aws now in fault state (train or signal doesn’t matter). C) Aws in fault doesn’t reset no longer functional. D) Second train passes previous chance for Aws to alarm. Driver confused as things are not happening as per training and only puts brake in when stopped train becomes unexpectedly visible.

Not impossible, but there’s a few issues with that.

RAIB do state that the issue with the Nottingham train was with its “on-train AWS”. Given how the RAIB do tend to me quite specific with their wording, this does seem to imply that the issue was with the train not the track, although I’m not sure we can infer than with certainty.

But secondly if the Corby train’s driver was approaching a red signal correctly and then did get the wrong warning at the magnet, or the AWS didn’t reset straight away, or whatever, one would still probably expect him to have stopped at the signal. Although not the most experienced of drivers, he doesn’t appear to have been a novice by any means.

And if he was flustered over something, just half a minute or so later he seems to have reacted promptly to seeing the train standing ahead.

It’s a really intriguing set of circumstances, and very interesting to discuss, however we just don’t have anything like enough info to be getting close to reaching an answer as to why.

 

[Signal Head]

Isn’t the junction signal approach controlled? Many are.

Yes, it is.

 

[dk1]

Yes, it is.

Thanks. I assumed it would be. It was stated shortly after the incident that flashing yellows did not apply here and obviously the speed difference made me think it had to be approach controlled.

 

[Banana]

Because WH356 was at Red. And it was at Red because a route could not be set to WH154 on the Up Fast, as the first train was still passing / occupying the section ahead up to WH154.

Thank you.

I understood that the train that was hit had been sitting on the Up Fast for some time, and didn't pass whilst the other train was stood in the station. Happy to be corrected.

To those going on about approach control, that would not apply to the station signal which would show an aspect one up from the signal controlling the junction.

 

[hexagon789]

And if he was flustered over something, just half a minute or so later he seems to have reacted promptly to seeing the train standing ahead.

Unless I've misinterpreted, the brake application that reduced the 360's speed from 76 to 49mph hasn't been confirmed as driver-initiated or by a safety system on the train such as the DVD/DSD (Vigilance/Driver Safety Device) or a failure to cancel the AWS or even an AWS fault.

 

[dk1]

Unless I've misinterpreted, the brake application that reduced the 360's speed from 76 to 49mph hasn't been confirmed as driver-initiated or by a safety system on the train such as the DVD/DSD (Vigilance/Driver Safety Device) or a failure to cancel the AWS or even an AWS fault.

It was on a right hand bend. To me it seems this is the point the driver realised that the 810 was in its path and applied the brake.

 

[Bald Rick]

Thank you.

I understood that the train that was hit had been sitting on the Up Fast for some time, and didn't pass whilst the other train was stood in the station. Happy to be corrected.

To those going on about approach control, that would not apply to the station signal which would show an aspect one up from the signal controlling the junction.

The Corby train departed Bedford Platfrom 1 at almost exactly the same time as the Nottingham train passed through at speed.

 

[bramling]

Unless I've misinterpreted, the brake application that reduced the 360's speed from 76 to 49mph hasn't been confirmed as driver-initiated or by a safety system on the train such as the DVD/DSD (Vigilance/Driver Safety Device) or a failure to cancel the AWS or even an AWS fault.

Very good point! I hadn’t clocked this. This does add another unknown, albeit one we should get an answer to fairly quickly.

 

[hexagon789]

It was on a right hand bend. To me it seems this is the point the driver realised that the 810 was in its path and applied the brake.

Very, very possibly, not saying that is not the case.

However, I read the RAIB interim report as not saying what caused the 360 to slow:

The OTDR from the rear vehicle of train 1H46 has been recovered and downloaded. While this OTDR does not record AWS activity, it does record the train’s speed and braking commands. Preliminary analysis of data from this OTDR shows that there was a brake activation around 9 seconds before the collision, while the train was travelling at approximately 76 mph (122 km/h). Preliminary analysis of OTDR data also indicates that the train was travelling at around 49 mph (79 km/h) at the point of collision.

That reads to me that the either haven't confirmed or are not yet stating what caused the "brake activation".

== Doublepost prevention - post automatically merged: 24 Jun 2026 ==

Very good point! I hadn’t clocked this. This does add another unknown, albeit one we should get an answer to fairly quickly.

If it was driver-initiated, I feel that "application" seems a more natural term to use than "activation" but regardless of semantics it's definitely not made explicit is it?

 

[yorkie]

But as a non train driver, what I'm struggling with is trying to imagine a set of circumstances where a speed of 76mph when the aspect of the previous 4 aspect signal was single amber is explicable. Can anyone help me with that? Can it only be as a result of presumption?

From speaking to drivers, and reading posts on this forum from drivers, these days it is made absolutely clear that no such presumption can be made, and this is especially so in the UK, where driving is noticeably more 'defensive' and 'cautious' than in many other countries I have visited.

However, people are human, so I am sure the investigation will look into whether this is a possibility of why the train initially accelerated, however as Haywain quite rightly said earlier...

I can see why, having been given the road to go across to the up fast, the driver might accelerate and think a normal clear road is ahead but that does not explain passing a red signal and not braking until the train ahead is seen.

There are all sorts of scenarious that could have occured, but it is best we wait for the next RAB update; perhaps more information on this may become available in an interim report, but I would guess that it may take some time.

 

[Banana]

The Corby train departed Bedford Platfrom 1 at almost exactly the same time as the Nottingham train passed through at speed.

Thank you.

 

[richa2002]

For me, one of the key factors is that the signal controlling passage over Bedford South Jn does not step up to the least restrictive aspect until really quite late. Not the AWS ramp like most of these setups but only 2 or 3 car lengths. From memory, I believe this is because there is another AWS ramp in the opposite direction on the same stretch of this bi-di piece of line for the signal controlling access to Bedford station. Presumably requiring this non-standard step up distance.

As a result, the driver really has to have a keen eye on what it steps up to and in my opinion, makes it more of a risk to just assume it's gone to green, like it will most of the time.

What feels unexplainable is then not reacting to the subsequent red. That next signal is visible well in advance and not on a curve which precludes early sighting. Even if the emergency brake was applied upon passing the AWS ramp for the red, the train would stop far short of the 810 in front. A total autopilot moment perhaps.

 

[Vince C]

It seems one possibility is the Corby train driver took the single yellow to be approach control for the crossover, not an indication that next signal is a 'red', hence his acceleration to 76mph once on the 'fast' which is not consistent with his expecting a 'Red' at the next signal. Please correct me if I'm wrong but I though approach control for turnouts were flashing yellows (single or double) to obviate that confusion. I have seen no mention that the single yellow at the crossover was flashing, so is my understanding incorrect?

 

[richa2002]

The signal for Bedford South Jn would be at red until the train is around the area of the overbridge north of this signal. As mentioned in my post, it then steps up to the least restrictive aspect at an unusually short distance from the signal, 2 or 3 car lengths from my estimation. There are no flashing yellows involved and if there were, these are at the signals that precede the crossover signal.

 

[CalderRail]

Very, very possibly, not saying that is not the case.

However, I read the RAIB interim report as not saying what caused the 360 to slow:



That reads to me that the either haven't confirmed or are not yet stating what caused the "brake activation".

== Doublepost prevention - post automatically merged: 24 Jun 2026 ==


If it was driver-initiated, I feel that "application" seems a more natural term to use than "activation" but regardless of semantics it's definitely not made explicit is it?

They did specifically say that they only, as yet, have the data from the recorder in the rear cab of the train, which only records the output/results of the train movements; getting to the one from the crumpled front cab, which will have recorded the driver's inputs, has taken them longer in the recovery process given the need to separate the trains, remove the corpse, record the positions of everything, and then go cutting through the wreckage to get at the recorder. So I think this is simply RAIB being their usual precise selves with their wording.

 

[yorkie]

It seems one possibility is the Corby train driver took the single yellow to be approach control for the crossover, not an indication that next signal is a 'red', hence his acceleration to 76mph once on the 'fast' which is not consistent with his expecting a 'Red' at the next signal. Please correct me if I'm wrong but I though approach control for turnouts were flashing yellows (single or double) to obviate that confusion. I have seen no mention that the single yellow at the crossover was flashing, so is my understanding incorrect?

Even if this was the case, I refer you to @Haywain's post above:

I can see why, having been given the road to go across to the up fast, the driver might accelerate and think a normal clear road is ahead but that does not explain passing a red signal and not braking until the train ahead is seen.