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

[fishwomp]

Isn't the problem here that rail vehicles are bouncing around at ~100mph?

No.

I'd assume they're running spinning disks, not flash memory for bulk storage with the age of the vehicles.

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.

Spinning disks certainly aren't going to like being constantly written to with the amount of vibrations and so on on a rail vehicle, so keeping a buffer in flash, then periodically purging to the bulk spinning disk would seem to me to have been sensible design when implemented?

The vibrations on a train should be no more than the vibrations for someone using a laptop on said train: they used spinning disks for long enough and still some today..

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..

We can / should be pretty sure the OTMR is recording all the time.. CCTV should be the same.

 

[Signal Head]

From this video:

I estimate signal WH152 comes into view about 700m on approach, just after passing over the A421 bridge. Mounted high up on the gantry, the green displayed would have likely reinforced 1H46's misconception they had movement authority until, coming around the long right-hand curve, the rear of 1B67 came into view a few coach lengths before the signal. This might help explain the very late braking. The loss of rear facing CCTV just before impact is interesting as it suggests a power loss, possibly an attempted reboot of all systems on the 810. If that was the case, did the rear tail lights extinguish? That would leave the stranded train rather invisible with no yellow ends or other prominent markings.

To add to the mix - WH356 approach release operates in two steps. R->Y occurs with the train some way back from the signal, while the further step up to less restrictive aspects is delayed until after the train has passed the AWS. So, this AWS operates as though 356 is a 'fixed Distant' for the signal ahead and does nothing to reinforce the aspect and hence acknowledgement of the conditions ahead.

Because of this, drivers may well be conditioned to cancel the AWS warning 'automatically', because they have to do it each time this movement is made irrespective of aspect received.

Some time back I had an opportunity to view some cab video for this movement, with a clear line ahead. WH356 stepped up to G approx 3 seconds before the train passed it (or perhaps before the signal passes out of view of a driver looking straight ahead - camera view is probably different to the human eye).

That's probably not a lot of time to register that this time the signal hasn't done what it 'always' does - step up to G, and of course there's no difference in the AWS indication received either.

This form of approach release is no longer permitted for new work and hasn't been for some time. I am aware of a more recent example where the controls were altered to release in two steps for sighting reasons, but in that instance there was a clear stipulation that any step up to G must take place before the train reaches the AWS, so that its usefulness is not diminished.

 

[WAO]

Just a gentle reminder that the Rule Book does not differentiate between track circuit and axle counters regarding the use of TCOCs when not able to contact the Signaller.

They MUST be placed down regardless.

The logic is that in a stressful situation, recalling the detection type is not reliable enough to decide whether to do so or not.

That's fine for train crew but does it help the panel?

What worries me is the 810 stopped underneath a clear signal, which, being higher up, might have been seen before the standing train appeared and so would be easy to "read through". A signal should be red if any part of its section is occupied.

Unless I have mis-read the report.

WAO

 

[AlterEgo]

The report doesn't appear to confirm whether the driver of the 360 was continuing to accelerate up to line speed at the point of the brake application or whether he had reached a target speed for double yellows and was maintaining it.

It will be interesting to note as he cancelled the AWS without, it seems, noticing the signal. Whether he believed it was green or double yellow will have an impact on the human factors involved here.

 

[cornishjohn]

I'd assume they're running spinning disks, not flash memory for bulk storage with the age of the vehicles.
Spinning disks certainly aren't going to like being constantly written to with the amount of vibrations and so on on a rail vehicle, so keeping a buffer in flash, then periodically purging to the bulk spinning disk would seem to me to have been sensible design when implemented?

Really? We have been using flash drives in the marine industries for over 20 years. Yes, decent ones cost a bit, but small beer compared to the price of rolling stock. The drives themselves are assembly replacements for spinning units, so interchangeable in ordinary loggers.

There is no reason to have more than a few seconds of video cached before writing to disk.

Other than poor choice of writing block size to storage page size, which is a bad way to try and save money.

 

[Annetts key]

What worries me is the 810 stopped underneath a clear signal, which, being higher up, might have been seen before the standing train appeared and so would be easy to "read through". A signal should be red if any part of its section is occupied.

Unless I have mis-read the report.

WAO

I think you misunderstand how automatic signals work. The signal that you are referring to was still showing green because the section ahead was clear. The section for a standard automatic signal normally starts approximately 200 yards (183 metres) to 300 yards beyond the signal (this is called the overlap) NOT at the signal.

This is perfectly safe in normal circumstances because the signal in rear (behind the train) is being held at red due to the track circuit the train is on being occupied.

In this particular case, the front of the train had not made it beyond the overlap hence why the automatic signal above it was still showing a green aspect. The report does give the distance to the overlap but as I'm on a mobile, I don't have the report to hand to quote from it.

If the driver of the stopping service train had obeyed the red signal, the collision would not have happened. It's the events that happened with that driver and their train that is the most important things that need the most attention.

 

[sharpener]

Just a gentle reminder that the Rule Book does not differentiate between track circuit and axle counters regarding the use of TCOCs when not able to contact the Signaller.

They MUST be placed down regardless.

The logic is that in a stressful situation, recalling the detection type is not reliable enough to decide whether to do so or not.

I can understand the logic of that in a way.

But it leads to a different problem. Once the clips are in place the crew may think they are safe(ish) from other traffic whereas unless they DO know which system is in place this may be a false sense of security.

ETA also if they are not in a track-circuited area they need to redouble their efforts to contact the signaller.

 

[stuving]

The relevant standard for CCTV systems (FF plus others, and including recorders and other elements as well as cameras) appears to be RIS-2712-RST issue 1.2 (2024). It says it contains requirements, but much of it is worded as suggestions - for example:

G 2.11.1.7 Consideration can be given to the use of solid state drives (SSD) instead of hard disk drives (HDD) as SSDs have better resistance to vibration due to the lack of internal moving parts. At the time of writing, initial costs of SSDs are considerably more than HDDs, however, the whole-life cost of using SSDs versus HDDs has been shown to be closer.

That wording is unchanged since issue 1 of 2020, and while it may have been true then about SSD purchase costs being higher it's now so close other factors should prevail.

Before that there was a guidance note: GM/GN2606 "Guidance on the Fitment and Functionality of Forward and Rear Facing Cameras on Rolling Stock" issue 1 2010. There was also a related standard: GMRT2472 "Requirements for Data Recorders on Trains" issue 1 2002, issue 2 2015 - on OTDRs but used as a reference in the guide when discussing crash-proof memory for CCTV (in 3.10.5).

This note talks about DVRs, and obviously expects hard drives to be used. It does recommend a video buffer, but for reasons I don't follow (someone with more knowledge of the technology at that date should do better).

What is installed in a train will obviously depend on when its design is this respect was committed, unless it has been updated. I imagine that the cameras and recorders in particular should be replaceable on a single-component basis where technical advances justify that. Obviously the 360 was built much earlier than the 810, though it may have been updated. Also it was running and its FF camera is important; for 1B67 it is the RF camera and it was not moving, and may have been undergoing some kind of reset, at the time.

All documents available from RSSB (https://www.rssb.co.uk/standards-catalogue). I'm currently having better success at finding a document there putting its ID code into Google than anything on their site!

 

[starlight73]

The report doesn't appear to confirm whether the driver of the 360 was continuing to accelerate up to line speed at the point of the brake application or whether he had reached a target speed for double yellows and was maintaining it. It will be interesting to note as he cancelled the AWS without, it seems, noticing the signal. Whether he believed it was green or double yellow will have an impact on the human factors involved here.

I’m not sure either. There is a period of 31 seconds where RAIB don’t say what happened with the train’s speed, unless I missed it (I’ve written this out and it’s between the bits marked in bold with a star * below).

Power was definitely applied at the beginning and end of these 31 seconds. The simple reading would be that acceleration/traction power application continued in these 31 seconds - but it’s not clearly stated whether it was sustained, or removed then reapplied.

The max speed of the train was 76mph, and it was 75mph when the emergency brake was applied (2 seconds after removing power).

• 17:12:46 — The class 360 was accelerating from 49mph and soon afterwards passed the AWS magnet
• *17:12:53 — this train passed WH154 at danger, (edited to add) 58mph still accelerating
• 17:13:13 — the class 810 became visible according to forward facing CCTV
• Time not stated — the class 360 reached a max speed of 76mph
• *17:13:24 — the driver reduced traction power demand
• 1 second later — full service brake
• 2 seconds later — emergency brake, 75mph
• 17:13:35 — collision, 50mph (stated in full report)

Page 13 of the report - bold added

53. At 17:12:46, train 1H46 received an AWS warning indication associated with signal WH154 (figure 5). FFCCTV images show that this signal was displaying a red aspect as the train approached. The OTDR from train 1H46 recorded that the driver acknowledged the AWS warning indication, but the train continued to accelerate towards and beyond the signal, which it passed at 17:12:53.

54. Train 1B67 first becomes visible in FFCCTV images from train 1H46 at 17:13:13. This footage ends approximately 2 seconds later. [...]

55. Analysis of the data from train 1H46’s OTDR shows that the driver reduced the traction power demand from 17:13:24 and then made a full-service brake application at 17:13:25. The OTDR recorded an emergency brake application at 17:13:27 when the train was travelling at around 75 mph (122 km/h). It is likely this sequence represents the driver of train 1H46 detecting the presence of train 1B67 on the line ahead and responding to it.

Page 19 of the report

80. OTDR data also shows that the driver of train 1H46 started to accelerate from around 49 mph (79 km/h) shortly before receiving the AWS warning indication for signal WH154, which was showing a red aspect. OTDR data shows that they acknowledged the AWS warning indication for this signal but continued to accelerate, reaching a maximum speed of 76 mph (122 km/h). The driver of train 1H46 initially applied the full service brake around 10 seconds before the collision, approximately 260 metres from the rear of train 1B67. About 2 seconds later, the emergency brake was applied, around 200 metres before the point of collision. The last available footage from train 1H46’s FFCCTV is around 600 metres from the rear of train 1B67. At this point, train 1B67 is visible but it is not possible from the footage to determine if the train is on the same line as train 1H46 due to track curvature. A train driver may experience the same limitation of visibility.

 

[WAO]

I think you misunderstand how automatic signals work. The signal that you are referring to was still showing green because the section ahead was clear. The section for a standard automatic signal normally starts approximately 200 yards (183 metres) to 300 yards beyond the signal (this is called the overlap) NOT at the signal.

This is perfectly safe in normal circumstances because the signal in rear (behind the train) is being held at red due to the track circuit the train is on being occupied.

I do understand this. My point is that it risks "reading through", a problem on long straights where multiple signals are in view.

Normal circumstances are not the best for judging safety systems. I am not speculating about the late driver's actions.

WAO

 

[Steve Harris]

I do understand this. My point is that it risks "reading through", a problem on long straights where multiple signals are in view.

Normal circumstances are not the best for judging safety systems. I am not speculating about the late driver's actions.

WAO

Which wasn't the case here.

 

[Farnborough]

The way I read @MarkyT's point was that, having passed WH154 at danger, the driver of 1H46 may have seen WH152 at green before seeing 1B67 stopped, and seeing that green may have reinforced the notion that 1B67 was racing ahead.

As a car driver, not a train driver, I know to stop at a red traffic light, even if the next one along the road is green...

And if I pass a red traffic light, I don't get a warning in the car - potentially I get a fixed-penalty notice in the post.

 

[pokemonsuper9]

As a car driver, not a train driver, I know to stop at a red traffic light, even if the next one along the road is green...

There are clearly enough car drivers that do read through that there are covers over those lights such that they can only be seen from certain areas.

 

[Annetts key]

I do understand this. My point is that it risks "reading through", a problem on long straights where multiple signals are in view.

Normal circumstances are not the best for judging safety systems. I am not speculating about the late driver's actions.

WAO

In my experience of investigating both "right side" and "wrong side" signalling failures, reading through on straight lines is rarely a problem if the drivers can see all the relevant signals in visual range. They get to know the route and know through experience how to interpret what they are seeing. Even where there is an aspect sequence that a lay person may not expect. And hence have no trouble obeying the correct signals.

That's one of the reasons that drivers should have good route knowledge.

Reading through becomes a real problem at night during darkness on curves where the human eye has difficulty working out the distance to signals that can only be seen for a very short amount of time. Especially if a signal that is not continuously visable (and not normally visable in daylight) can be seen just once or more than once, but very intermittently through whatever obstruction (often trees or other vegetation) there are in the locality.

Where the signal ahead in the distance that can only be seen intermittently is showing a proceed aspect (typically a green aspect) but the next signal that actually applies to the train is showing a more restrictive aspect, this has caused major problems in the past at certain locations.

In some cases, special controls can be provided to hold the "troublesome" signal at red unless a train is routed up to it or the signalling defects that a train is actually approaching it.

Or it may be possible to reposition the signal.

 

[AlterEgo]

As a car driver, not a train driver, I know to stop at a red traffic light, even if the next one along the road is green...

This isn't an enlightening comment. Read-through is a known phenomenon with repetitive train driving work.

 

[tram21]

This isn't an enlightening comment. Read-through is a known phenomenon with repetitive train driving work.

Its also something that happens to many car drivers too when there are multiple traffic lights!

 

[bramling]

That's fine for train crew but does it help the panel?

What worries me is the 810 stopped underneath a clear signal, which, being higher up, might have been seen before the standing train appeared and so would be easy to "read through". A signal should be red if any part of its section is occupied.

Unless I have mis-read the report.

WAO

This might have been a factor in the driver taking a few seconds to react, but it doesn’t explain the real cause, which was the complete disregard of the red aspect. In fact, I wouldn’t necessarily say that the driver was particularly slow in reacting to the train standing ahead - by the time one allows a few seconds form him to have seen it, processed what he was seeing, probably done a double take as the horror of the situation dawned on him, I’d say the reaction time is probably what we would expect to have to seen. ISTR the Ladbroke Grove driver also took a few seconds to process what was unfolding.

There’s no doubt a massive amount to be unpicked on the human factors. However unless something fairly obvious comes to light, it’s one of those mysteries which may never be solved.

 

[norbitonflyer]

I do understand this. My point is that it risks "reading through", a problem on long straights where multiple signals are in view.


WAO

The possibility of "reading through" would have existed whheher or not the front of the train had passed the signal itself. The fact it had done so merely meant the point of collision was a few yards further on than if it had stopped before reaching the signal.

I recall in "red for danger" it used to be a rule (and may still be) that if you have passed a signal at danger (whether on the authority of a signalman or otherwise) you should not assume that the line beyond the next signal is clear, whatever aspect it is showing - it may not yet have been restored to danger after the train ahead of you.

 

[Krokodil]

As a car driver, not a train driver, I know to stop at a red traffic light, even if the next one along the road is green...

You're not doing 110mph.

 

[Signal Head]

As a car driver, not a train driver, I know to stop at a red traffic light, even if the next one along the road is green...

And if I pass a red traffic light, I don't get a warning in the car - potentially I get a fixed-penalty notice in the post.

I don't think he would have been able to see both signals at the same time owing to the curvature, so it's not 'reading through' the Red - more like missing the Red (for whatever reason(s)), then saw the Green before the rear of the train in front. In other words it wasn't the visibility of the Green which caused the Red to be missed/overlooked.

 

[AlterEgo]

I don't think he would have been able to see both signals at the same time owing to the curvature, so it's not 'reading through' the Red - more like missing the Red (for whatever reason(s)), then saw the Green before the rear of the train in front. In other words it wasn't the visibility of the Green which caused the Red to be missed/overlooked.

This is why the acceleration profile of the impact train will be important, and I was surprised to see it omitted from the interim report as it must surely be known. Was he driving like he was on greens or double yellows?

 

[bramling]

As a car driver, not a train driver, I know to stop at a red traffic light, even if the next one along the road is green...

I don’t reckon there’s any car driver who hasn’t come close to accidentally reading through a red light at some point.

I think this focus on “read through” is a bit of a distraction though. The accident couldn’t have been caused by a read through, as the next signal wasn’t visible from the one passed. If anything read through might have given us an explanation, had it been plausible.

Unfortunately, the reason for this accident could actually be something as simple as a momentary distraction just at the wrong moment. No doubt any mobile phone records will have been examined. What’s more unusual is the complete lack of response to the red though, more normal in these sorts of situations is a sudden response at the last moment.

 

[Signal Head]

This is why the acceleration profile of the impact train will be important, and I was surprised to see it omitted from the interim report as it must surely be known. Was he driving like he was on greens or double yellows?

In an earlier post I mentioned having seen some cab video. That had speed data attached and either at or shortly after (can't remember which) WH154 the speed was 76MPH (running on Green aspects from the junction signal onwards). I think it's 50 through the junction, acceleration started after an allowance for the rear end clearing the turnouts. So 76 for the incident 360 at or near the same point suggests it was being driven as if on Greens.

 

[DadimF]

As a car driver, not a train driver, I know to stop at a red traffic light, even if the next one along the road is green...

And if I pass a red traffic light, I don't get a warning in the car - potentially I get a fixed-penalty notice in the post.

Road traffic signals and rail signals can never be used as a comparison.

Without talking about this situation specifically, signals mainly at junctions can have controls which hold the signal at red until the train reaches a defined point where it will step up to a Y or G dependent on the condition of the route ahead. These controls are used to control the speed of the train via driver behaviour in response to the aspect shown.

Now if you approach those signals all the time for months and 1st signal goes R>Y, then the next signal always showed Y or G, the driver can become subject to almost working in an automatic way because the expectation of seeing that aspect sequence has been reinforced for so long.

Now if one time out of a hundred the aspects instead showed R>Y like normal, but the second signal R, that repetitive physical action of pressing the AWS as usual plus almost unconsciously expecting a Y can cause a mental blindness.

This is a well known phenomenon that means you could be the best, safest driver ever, but you can't overcome being a human with a brain that can be subject to these kinds of things unwillingly. And no driver should blamed when it happens to them because it's pretty much never malicious, it's the system providing the automatic programming behaviours that needs to be rectified.

 

[Killingworth]

Road traffic signals and rail signals can never be used as a comparison.

Without talking about this situation specifically, signals mainly at junctions can have controls which hold the signal at red until the train reaches a defined point where it will step up to a Y or G dependent on the condition of the route ahead. These controls are used to control the speed of the train via driver behaviour in response to the aspect shown.

Now if you approach those signals all the time for months and 1st signal goes R>Y, then the next signal always showed Y or G, the driver can become subject to almost working in an automatic way because the expectation of seeing that aspect sequence has been reinforced for so long.

Now if one time out of a hundred the aspects instead showed R>Y like normal, but the second signal R, that repetitive physical action of pressing the AWS as usual plus almost unconsciously expecting a Y can cause a mental blindness.

This is a well known phenomenon that means you could be the best, safest driver ever, but you can't overcome being a human with a brain that can be subject to these kinds of things unwillingly. And no driver should blamed when it happens to them because it's pretty much never malicious, it's the system providing the automatic programming behaviours that needs to be rectified.

I have noted two SPADS at signals near where I live where that may have been a factor. That said both drivers did realise in time that the red wasn't going to change and came to a halt not far beyond the signals but far enough to have to be relieved delaying trains for the better part of 2 hours.

In this case it seems the 810 was brought automatically to a halt at a green signal but the 360 wasn't at a red.

 

[DadimF]

I have noted two SPADS at signals near where I live where that may have been a factor. That said both drivers did realise in time that the red wasn't going to change and came to a halt not far beyond the signals but far enough to have to be relieved delaying trains for the better part of 2 hours.

In this case it seems the 810 was brought automatically to a halt at a green signal but the 360 wasn't at a red.

That's classic examples of anticipating the signal to always step up at the exact same spot and being ready to accelerate like they always have, but luckily in these cases their brains registered the red remaining on though too late to stop prior.

But it only takes one of those situations to happen to a driver suffering from being in a more complacent and automatic state of mind to turn into a disaster.

There's been quite a lot of talk about these situations recently and how we can mitigate it but unfortunately the nature of the existing layout and alterations required for that specific layout would usually be so expensive we either leave it alone or try to design in something to provide a minor betterment where the low cost is justifiable. Even the tiniest simple alteration on data based interlocking controls that would take literally minutes to do costs upwards of 100k so we as signalling designers are constantly limited in what we can design which is frustrating. It all comes down to money like everything else.

 

[leezer3]

The vibrations on a train should be no more than the vibrations for someone using a laptop on said train: they used spinning disks for long enough and still some today.

Totally different duty cycles...

Using a laptop will be periodic disk accesses, followed by the disk parking the heads, maybe for an hour or so a day. The disk controller will also perform aggressive head parking if it detects high vibration.

Really? We have been using flash drives in the marine industries for over 20 years. Yes, decent ones cost a bit, but small beer compared to the price of rolling stock. The drives themselves are assembly replacements for spinning units, so interchangeable in ordinary loggers.

I think you're forgetting the inherant conservatism of the railway in general, and to a certain degree the massive unionisation 'issues'- Any change to a system, especially one with potential safety implications or interfaces has to go through a massive hoopla of change validation, union acceptance etc. etc. even when it's clearly beneficial.


The safety, and the related attitudes of the railway is far closer to that of aviation, and that's why even if the underlying storage has been swapped for flash it's retained the limitations of the old drives.

 

[norbitonflyer]

I have noted two SPADS at signals near where I live where that may have been a factor. That said both drivers did realise in time that the red wasn't going to change and came to a halt not far beyond the signals

That's classic examples of anticipating the signal to always step up at the exact same spot and being ready to accelerate like they always have, but luckily in these cases their brains registered the red remaining on though too late to stop prior.

Isn't this what happened at Colwich? Although in that case the train came to a stand across the junction the red was protecting.

I think this focus on “read through” is a bit of a distraction though. The accident couldn’t have been caused by a read through, as the next signal wasn’t visible from the one passed. If anything read through might have given us an explanation, had it been plausible.

The green may not have been visible from the previous signal, but if the driver believed he had passed the prevous signal showing a proceed aspect, the sight of the green signal ahead would have reinforced that belief.

In some places a separate "overlap" track circuit is used just beyond a signal, causing both that signal and the preceding one to be held at red when it is occupied. (It was the occupation of such a circuit at Clapham Junction in 1988 that caused the false green to briefly show red as the next train approached, causing it to stop) Had such an overlap track circuit been provided at Elstow, the 810 would probably have occupied it and made the signal it was standing under go red. Of course, it would not have helped if then train had stopped short of the signal, and it is unlikely in any case that such provision would be a priority on a plain line section like Elstow (any mkore than TPWS was seen as worth fitting there)

 

[Taunton]

To add to the mix - WH356 approach release operates in two steps. R->Y occurs with the train some way back from the signal, while the further step up to less restrictive aspects is delayed until after the train has passed the AWS. So, this AWS operates as though 356 is a 'fixed Distant' for the signal ahead and does nothing to reinforce the aspect and hence acknowledgement of the conditions ahead.

Because of this, drivers may well be conditioned to cancel the AWS warning 'automatically', because they have to do it each time this movement is made irrespective of aspect received.

Some time back I had an opportunity to view some cab video for this movement, with a clear line ahead. WH356 stepped up to G approx 3 seconds before the train passed it (or perhaps before the signal passes out of view of a driver looking straight ahead - camera view is probably different to the human eye).

That's probably not a lot of time to register that this time the signal hasn't done what it 'always' does - step up to G, and of course there's no difference in the AWS indication received either.

This form of approach release is no longer permitted for new work and hasn't been for some time. I am aware of a more recent example where the controls were altered to release in two steps for sighting reasons, but in that instance there was a clear stipulation that any step up to G must take place before the train reaches the AWS, so that its usefulness is not diminished.

This is an interesting comment for the enquiry to consider. Stepping up signals just a few seconds before the train reaches them was the key factor in the 1995 Toronto Subway major accident, which if this is actually the case with WH356 it would be instructive for the Bedford enquiry to read about for parallels - for speed control purposes signals were clearing on a timer, and drivers got used to this over time and would approach red signals at speed, with them clearing just in time, sometimes just as they passed from view. I saw this myself from the train front windows before the accident and was surprised at the approach, and an article in a UK rail magazine made similar comments.

Protection for the section ahead was by trip train stops, like the London Underground, and indeed they had periodic trips. In the Toronto accident the train ahead was unusually delayed (sound familiar from Bedford?), but the trip mechanism was out of alignment. Here's a summary ...

1995 Russell Hill subway accident - Wikipedia

en.wikipedia.org

 

[Purple Train]

In an earlier post I mentioned having seen some cab video. That had speed data attached and either at or shortly after (can't remember which) WH154 the speed was 76MPH (running on Green aspects from the junction signal onwards). I think it's 50 through the junction, acceleration started after an allowance for the rear end clearing the turnouts. So 76 for the incident 360 at or near the same point suggests it was being driven as if on Greens.

The RAIB mentions the train as having passed WH154 at 58mph while accelerating. The maximum speed of 76mph was reached 31 seconds later. Would that be typical acceleration for a 360 constantly on full power?