Collision and derailment near Salisbury (Fisherton Tunnel) 31/10/21

[BayPaul]

If you read the BTP statement it says the GWR hit the SWR unit.

That statement also says the front few carriages remained upright while the back tipped on their side. Unless I am completely misunderstanding the pictures, that seems to be inaccurate - the rear carriages of both trains, outside the tunnel are upright, the front of the SWR train inside the tunnel is at an angle, so perhaps it should not be treated as a source of gospel truth either.

 

[Moriarty01]

Christian Wolmar was on the radio earlier giving his opinion/speculation on the incident and he raised the point that the second train may have had a red signal but was unable to brake in time due to leaves on the lines / general slippage causing it to skid into the first train.

 

[43096]

I don't think it's possible to draw that conclusion for certain. The timing of the second movement so close behind the first suggests a signalled move. A verbal proceed past a red would probably have taken longer, and the protecting signal is also TPWS fitted, so a SPAD should have resulted in emergency braking which would have brought the train to a stand before the junction conflict point.

And what if the train was in emergency braking and it slid past the signal due to poor railhead conditions and into collision with the GWR service.

It appears that the railhead treatment train on the Basingstoke-Salisbury line didn’t run yesterday.

 

[Whistler40145]

Christian Wolmar was on the radio earlier giving his opinion/speculation on the incident and he raised the point that the second train may have had a red signal but was unable to brake in time due to leaves on the lines / general slippage causing it to skid into the first train.

Do you mean that the train coming from the Romsey direction was probably slowing for the Junction towards Salisbury, but was slipping on wet leaves and couldn't brake in time and shot through a red signal?

 

[Efini92]

The lesson of not legging it down the line to protect the train was learnt, with tragic consequences, in the 1995 Ais Gill accident. It may well also turn out to be the case that the subject of emergency protection may be something of relevance during last year's Carmont accident (where there was certainly considerable scope for a secondary collision). It seems unlikely to have been a salient factor yesterday though.

Agreed. It’s possible the 158 came to a stand over the junction and the driver was trying to fault find.

 

[Scotrail84]

Christian Wolmar was on the radio earlier giving his opinion/speculation on the incident and he raised the point that the second train may have had a red signal but was unable to brake in time due to leaves on the lines / general slippage causing it to skid into the first train.

Even if that were the case the driver would have most likely reacted to previous cautionary aspects bringing the train speed down. I don't know the area but it would seem unlikely that the following service slipped for a long time at high speeds into the back of the stationary train although I don't want to speculate too much on the matter.

 

[bramling]

I don't think it's possible to draw that conclusion for certain. The timing of the second movement so close behind the first suggests a signalled move. A verbal proceed past a red would probably have taken longer, and the protecting signal is also TPWS fitted, so a SPAD should have resulted in emergency braking which would have brought the train to a stand before the junction conflict point.

TPWS isn't guaranteed to stop a train within the overlap, so I don't think this necessarily holds true.

A number of theories have been put forward for how any potential SPAD might have occurred, but I think it's best to leave it there on that score for now, as that will be a very complex subject, and even those investigating won't have the full picture yet.

 

[Peter Sarf]

Agreed. It’s possible the 158 came to a stand over the junction and the driver was trying to fault find.

Does that mean the 158 was due to enter the junction AFTER the 159 ?.
What does the timetable say ?.

 

[Whistler40145]

I wonder when the area had a visit from the RHTT?

 

[Ianno87]

Does that mean the 158 was due to enter the junction AFTER the 159 ?.
What does the timetable say ?.

Pretty sure in the standard hour, the Waterloo-Exeter usually follows the Portsmouth-Bristol, which gives a good connection for passengers doing journeys like Southampton-Exeter.

Although the timetable was heavily disrupted yesterday, so the planned sequence may not have happened.

 

[bramling]

What is really tragic about Ais Gill is that the one fatality appears to have been the person who was doing the most to protect the train and its passengers.

Depends how one looks at it. At Ais Gill the guard was certainly looking after the passengers, however it was his responsibility to have protected his train, which wasn't done. On an absolute block railway this was of critical importance, and had the guard been carrying out the protection duties then he almost certainly wouldn't have succumbed to the secondary collision. All there is to say is this was a very tragic outcome indeed.

 

[backontrack]

Depends how one looks at it. At Ais Gill the guard was certainly looking after the passengers, however it was his responsibility to have protected his train, which wasn't done. On an absolute block railway this was of critical importance, and had the guard been carrying out the protection duties then he almost certainly wouldn't have succumbed to the secondary collision. All there is to say is this was a very tragic outcome indeed.

Certainly true, although the Crewe control centre did say "we'll take care of that, driver". But yes, it was very tragic and certainly avoidable.

 

[Greybeard33]

Times at Tunnel junction forthe two GWR trains according to Real Time trains were:
1F27 1841.75
1F30 1842.75,
so they would have passed each other on the curve between Tunnel Junction and Laverstock South.
The last recorded time for 1L53 is at Laverstock North, at 1842.5, a few seconds before 1F30 passed Tunnel Junction and entered the tunnel itself.

The scheduled time from Laverstock North Jn to Tunnel Jn is 0.5 minutes according to RTT.

If the above timings were accurate, and if the SWR train passed Laverstock North Jn at line speed, it would have arrived at Tunnel Jn at 1843, approximately 0.25 minutes after the GWR train was timed at the junction.

 

[Annetts key]

My fear is that if sufficient of the first train had left the track in order for the track circuit to clear, then the automatic route setting could have stepped in to set the route for the following train immediately before anyone had any time to think, and because the track circuit was still showing clear, the signal would have cleared straight away as commanded.

Do we know if there is point to point locking between points 105 and 106?

Things we know:
A GWR train (1F27 16:58 Bristol Parkway to Portsmouth Harbour) had left Salisbury at about 18:37 on the Up Main just before the incident. Going through the tunnel and crossing over the junction onto the Up Dean, but it stopped before it got to Dean.

GWR train (1F30 17:08 Portsmouth Harbour to Bristol Temple Meads) passed Dean (on the Down Dean line) at about 13:35, and approached the junction (Tunnel Junction), it was either going slowly or for some reason stopped over the junction on the Down Main, with the rear part still on the junction, but with the front part in the tunnel.

If this train stopped before the collision, we do not know why this train stopped, but drivers don’t stop their trains in a such a location without there being a good reason.

SWT train (1L53 1720 London Waterloo to Honiton) was in the area of Andover at 18:29 on the Down Main, it continued on towards Tunnel Junction and hit GWR train 1F30 at the junction where the Down Main and Down Dean converge. The collision resulted in it derailing and some of the units ending up at an angle. The leading cab suffering extensive damage.

The public live train websites (which obtain their feed via a Network Rail system that in turn gets some of its information from the various Train Describer systems) show the ‘head code’ (TD ‘number’) for 1L53 still in the ‘berth’ for signal SY31.

Questions (I not looking for answers, just things to think about, the kinds of questions that the investigators will be asking, so please don’t speculate here):

What was the reason that caused the driver of GWR train 1F30 to stop their train where they did? Did it actually hit something? Was there a problem with the points fittings (point number 106) ? Or was there a problem with the train? If this train did hit something or if there was a problem with 106 points, did this train derail? Or was it only derailed the two trains collided?

Why did the TD not step the ‘head code’ forward for 1L53 when the train has clearly passed signal SY31?

If there was no problem with GWR train 1F30 operating the track circuits (where it is in the photos I would expect it to be on at least two track circuits) the interlocking should have held signal SY31 at red. So was there a problem with the track circuits?

In the photos, points 105 look to be lying reverse, as they would have been needed for the move of 1F27 over the junction. This is the opposite way to that required for 1L53 if there is point to point locking between points 105 and 106. Even if there is no point to point locking, 105 points should be required to be normal for flank protection.

In which case then if points 105 are not set correctly, signal SY31 should not clear. Was there a problem here?

We don’t know which way points 106 are because we can’t see them and they are likely to be under the GWR train. If 106 points were in the correct position for the GWR train 1F30, then they can’t be in the correct position for the SWT train 1L53. So if 106 points are not detected to be in the correct position, then signal SY31 should not clear. Was there a problem here?

Did the signaller (or ARS) route signal SY31? If the answer is no, then signal SY31 should not clear to a proceed aspect. Did signal SY31 stay red? Was it lit? What was the sighting like? Could approaching train drives see it clearly and at the required minimum distance?

If SY31 does not show an proceed aspect or if the signal is not illuminated , then the AWS should result in the AWS horn sounding in the leading cab of train 1L53. Was the track and lineside equipment for the AWS working? Was the train AWS equipment working?

SY31 should be fitted with TPWS equipment. The TPWS transmitter loops should be energised at all times unless the signal is showing a proceed aspect. Was the track and lineside equipment for the TPWS working? Was the train TPWS equipment working?

The signal, SY31 requires electricity to work, as does the TPWS. Was there a loss of power to this equipment, causing SY31 to go ‘black’ and disappear into the darkness? The signal identification plate/sign should be made of the reflective type so that it can be seen more easily in poor light conditions. Was it? Was it clean?

If there was a loss of power due to a fault with the signalling system, what caused this?

If there were no faults with the signalling system, was there a problem with the braking system on train 1L53? Or was there rail head contamination that reduced the effectiveness of the braking system?

I’m sure there are many other questions. As I said earlier, please don’t speculate. The facts will come out in the report once the investigation is complete.

Edited on Monday 1st at 21:08 to update post with corrections and new information.
Further edited on Monday 1st at 21:47 to correct a typo (1K53 instead of 1L53).

 

[pompeyfan]

A lot of the passengers interviewed in the media described juddering and vibration shortly before the collision. That sounds suspiciously like WSP activity.

 

[Horizon22]

Do we know if there is point to point locking between points 105 and 106?

Things we know:
A GWR train (1F27 16:58 Bristol Parkway to Portsmouth Harbour) had left Salisbury at about 18:37 on the Up Main just before the incident. Going through the tunnel and crossing over the junction onto the Up Dean, but it stopped before it got to Dean.

GWR train (1F30 17:08 Portsmouth Harbour to Bristol Temple Meads) passed Dean (on the Down Dean line) at about 13:35, and approached the junction (Tunnel Junction), but for some reason stopped over the junction on the Down Main, with the rear part still on the junction, but with the front part in the tunnel. The front part does not look to be derailed in the photos.

We do not know why this train stopped, but drivers don’t stop their trains in a such a location without there being a good reason.

SWT train (1L53 1720 London Waterloo to Honiton) was in the area of Andover at 18:29 on the Down Main, it continued on towards Tunnel Junction and hit GWR train 1F30 at the junction where the Down Main and Down Dean converge. The collision resulted in it derailing and some of the units ending up at an angle. The leading cab suffering extensive damage.

The public live train websites (which obtain their feed via a Network Rail system that in turn gets some of its information from the various Train Describer systems) show the ‘head code’ (TD ‘number’) for 1L53 still in the ‘berth’ for signal SY31.

Questions (I not looking for answers, just things to think about, the kinds of questions that the investigators will be asking, so please don’t speculate here):

What was the reason that caused the driver of GWR train 1F30 to stop their train where they did? Did it actually hit something? Was there a problem with the points fittings (point number 106) ? Or was there a problem with the train? If this train did hit something or if there was a problem with 106 points, did this train derail?

Why did the TD not step the ‘head code’ forward for 1L53 when the train has clearly passed signal SY31?

If there was no problem with GWR train 1F30 operating the track circuits (where it is in the photos I would expect it to be on at least two track circuits) the interlocking should have held signal SY31 at red. So was there a problem with the track circuits?

In the photos, points 105 look to be lying reverse, as they would have been needed for the move of 1F27 over the junction. This is the opposite way to that required for 1L53. If points 105 are not set correctly, signal SY31 should not clear. Was there a problem here?

Did the signaller route signal SY31? If the answer is no, then signal SY31 should not clear to a proceed aspect. Did signal SY31 stay red? Was it lit? What was the sighting like? Could approaching train drives see it clearly and at the required minimum distance?

If SY31 does not show an proceed aspect or if the signal is not illuminated , then the AWS should result in the AWS horn sounding in the leading cab of train 1L53. Was the track and lineside equipment for the AWS working? Was the train AWS equipment working?

SY31 should be fitted with TPWS equipment. The TPWS transmitter loops should be energised at all times unless the signal is showing a proceed aspect. Was the track and lineside equipment for the TPWS working? Was the train TPWS equipment working?

The signal, SY31 requires electricity to work, as does the TPWS. Was there a loss of power to this equipment, causing SY31 to go ‘black’ and disappear into the darkness? The signal identification plate/sign should be made of the reflective type so that it can be seen more easily in poor light conditions. Was it? Was it clean?

If there was a loss of power due to a fault with the signalling system, what caused this?

This is a very clear and thought-through question list.

There's definitely key points nobody yet knows (or lots of people know key pieces of information but nobody yet knows the whole) which is exactly why the RAIB get involved and comb through everything and get a clear sequence of events and the "why" in addition to the "what" that has happened. As always there will be the human factors questions too - Why did the train driver make the decision at that given time? What was the signaller's process and why? When was the alarm first raised? Were all processes followed? And so on. These will be confirmed by tape reads at the box and the data recorders on the train.

 

[the sniper]

Christian Wolmar was on the radio earlier giving his opinion/speculation on the incident and he raised the point that the second train may have had a red signal but was unable to brake in time due to leaves on the lines / general slippage causing it to skid into the first train.

Last night an older female passenger was interviewed by BBC News, she talked about leaves on the line, which seemed a bit random at the time.

 

[BayPaul]

Things we know:

GWR train (1F30 17:08 Portsmouth Harbour to Bristol Temple Meads) passed Dean (on the Down Dean line) at about 13:35, and approached the junction (Tunnel Junction), but for some reason stopped over the junction on the Down Main, with the rear part still on the junction, but with the front part in the tunnel. The front part does not look to be derailed in the photos.

We do not know why this train stopped, but drivers don’t stop their trains in a such a location without there being a good reason.

I don't think we do know this. The last few posts, and some recent official statements imply that the GWR train may not have stopped until the SWR train hit it. Accurate or not, I don't think it can be listed for certain that the GWR train was stopped.

 

[MarkyT]

TPWS isn't guaranteed to stop a train within the overlap, so I don't think this necessarily holds true.

Sometimes the junction conflict point is immediately ahead of the end of the overlap, sometimes it is further away. At the speeds involved here, standard overspeed and trainstop fitment should be able to bring most* trains to a stand before the conflict. At junctions with higher approach speeds where a standard fitment would not provide full protection, additional overspeed installations are provided on approach to the signal (TPWS+).

*There is some concern that certain freight train formations may not always be able to fully stop in time, although their speed would nevertheless be significantly reduced. Not relevant in this case, clearly.

 

[Dan17H]

I'm due there tomorrow as part of the Structures Team investigating the part any local structures would/could of had on the incident. I'm sure the RAIB alongside the BTP will release an interim report shortly after that.

 

[Ianno87]

I don't think we do know this. The last few posts, and some recent official statements imply that the GWR train may not have stopped until the SWR train hit it. Accurate or not, I don't think it can be listed for certain that the GWR train was stopped.

Partly because it would be such an unusual place for the GWR train to have stopped in the first place. So I agree, not definitive that it had stopped prior to the collision.

As I said upthread,.the only present certainty is the final positions of the trains. Everything else is, at present, speculation.

 

[bramling]

Certainly true, although the Crewe control centre did say "we'll take care of that, driver". But yes, it was very tragic and certainly avoidable.

Unfortunately in the case of Ais Gill, control's words were meaningless, as protecting the train wasn't in their gift. Equally, with the best will in the world the guard shouldn't have taken any assurance from this, especially as "we'll take care of that" isn't the same as "I can confirm your train is protected and there's nothing else in the section". Whilst Ais Gill was not an accident which has found itself particular etched in history, the lesson learned is one every single train crew anywhere should never forget. It's rather unfortunate that we saw further sloppiness from control just a couple of years later at Southall, with a failure to grasp the seriousness of AWS being isolated. Again, something never to escape the mind. Almost certainly not relevant yesterday however, where it seems everything happened very quickly (I'm emphasising this in case anyone might read this and jump to inaccurate conclusions).

 

[Annetts key]

Post #465 corrected.

 

[Andyh82]

"Passengers fearing they were about to die had called loved ones to say goodbye as investigators probe how a busy locomotive was a 'sitting duck' for seven minutes before a second high-speed service smashed into it when signals failed to turn red. "

Eh?

Indeed
That quote has gone from overly dramatic to by the sounds of it a complete fabrication?

 

[462cd]

There are quite a few interviews now from passengers who were involved. Has anyone heard a single witness who describes their train being stopped at the time of impact?

 

[Snow1964]

A lot of the passengers interviewed in the media described juddering and vibration shortly before the collision. That sounds suspiciously like WSP activity.

That suggests anti lock braking struggling to avoid wheels locking and slipping.

It is quite clear from the photographs that vegetation (trees with big leaves) are in the cutting, and there are plenty of freshly fallen leaves on the ballast.

It is also known from weather reports that it had been raining on the day, so leaves would be wet.

It could simply be the signal was red, but train skidded past, which is consistent with colliding with side of other train.

However I don’t want to speculate on if trees were out of control, or if rail head treatment trains were insufficient to avoid slippery rails approaching the junction.

 

[norbitonflyer]

Do you mean that the train coming from the Romsey direction was probably slowing for the Junction towards Salisbury, but was slipping on wet leaves and couldn't brake in time and shot through a red signal?

I think he meant the London train, not the Romsey train. The outcome would have been similar if either over-ran the junction whilst the other had the road, but given another GWR train was crossing the junction on its way towards Romsey less than a minute before, the London train could (or at least should) not have been signalled across the junction at the same time.

 

[swt_passenger]

D

Indeed
That quote has gone from overly dramatic to by the sounds of it a complete fabrication?

IMHO its looking like completely made up. Unfortunately that regurgitated 7 minute delay leads to discussions going off at tangents about protection of the line, and completely different scenarios like Ais Gill etc…

 

[QueensCurve]

Wow, that looks bad. Lucky escape for all on board. Also very lucky it didn't result in fuel everywhere and the possible effects of that in a tunnel which don't really bear thinking about.

In some ways not dissimilar to the collision in the tunnel north of Watford a few years ago?

The Watford collision of 16 September 2016 involved a train derailing due to an obstruction (landslip) and then colliding with an oncoming train.

 

[bramling]

D

IMHO its looking like completely made up. Unfortunately that regurgitated 7 minute delay leads to discussions going off at tangents about protection of the line, and completely different scenarios like Ais Gill etc…

The 7 minutes could well refer to something else. Given the circumstances of what happened, it's quite possible - for example - there could have been seven minutes before any reassurance from staff, at least in some parts of the trains. The seven minutes may well have come from somewhere relevant, but it doesn't seem to be a causal factor in the incident, as we have more reliable info showing timings. I say "more reliable" with the caveat that most of this data comes from open source data, but I'd say it's of value precisely because it's raw data, devoid of anyone's opinion.