Great Western Coffee Shop

Well what they hoped wouldn't happen has happened, an incident outside Waterloo station involving an engineering train and a passenger train having a coming together at a set of points.

SWT are saying 'avoid Waterloo'.

Judging by the pictures on Paul Clifton's Twitter feed there could be major disruption for quite sometime:
https://twitter.com/PaulCliftonBBC

From the BBC http://www.bbc.co.uk/news/uk-england-london-40933704
Train derails outside London Waterloo station
5 minutes ago
Three people were checked over by London Ambulance Service but no-one was hurt
Passengers have been advised to avoid London Waterloo station all day after a train derailed outside the station.
The train partly left the tracks as it was pulling away from a platform at low speed at 05:40 BST, Network Rail said.
Three people were checked over by London Ambulance Service but nobody was taken to hospital.
Ten platforms are already shut as part of engineering works to enlarge the station's capacity.
The station is usually one of the busiest in the UK
Passengers were already affected by ongoing engineering works
Waterloo works: What you wanted to know
Emergency services including police, paramedics and fire crews were called to the scene.
Photos shared by passengers on social media suggest the passenger train had struck an engineering wagon.
Chris Denham, a spokesman for Network Rail, told the BBC the partially derailed train had blocked three of the platforms so only very limited services could run.
"If you don't have to travel to Waterloo today, please don't," he said.
Passengers are being advised to use alternative routes
South West Trains said there would be a very limited service running to and from Waterloo "which is expected for the rest of today".
Many of the trains running have been cancelled or are are stopping or starting short of the station, the rail operator said.
Passengers are being advised to use alternative routes and have been warned other stations in the area will be "very busy" with the possibility of queuing systems being put in place.
The ongoing £800m engineering project at Waterloo will prepare the station for longer trains and provide space for 30% extra passengers during the busiest times of the day.
The station is usually one of the busiest in the UK, with an average of 270,000 journeys made to and from it each day.

There appears to be some diversions of SWT West of England services into Reading. I've just spotted on Open Train Times a 1Z26 which has gone via Mortimer into P3 at RDG. It's not showing up on RTT but on LDB there's a 0945 arrival at Reading from Exeter St David's via Salisbury and Basingstoke. The next departure that way showing on LDB is a 1002 from RDG to Salisbury. That's then followed by a 1011 arrival from EXD departing back there at 1047. None of these currently appear on RTT at RDG.

Back to Waterloo.  You can see the position of the engineering train wagons and the derailed train here: http://sentry.og2.co.uk/images/ICE00012/2017-08-15/ICE00012-2017-08-15_102200.jpg

(http://sentry.og2.co.uk/images/ICE00012/2017-08-15/ICE00012-2017-08-15_102200.jpg)

Yes according to Journeycheck all Exeter to Waterloo services are being diverted to/from Reading. The Benefits of diesel powered trains.

That engineering train is in P10, isn't it? The very one that (with its approach line(s)) was made part of the possession by a late change in July? So, if there was something in the plans or instructions to signallers that wasn't changed then, so a train leaving P11 could use the line to P10 ...

AIUI the engineering train has been in that position every day since the work started.  It ought to be impossible to set a route onto the line it is standing on.  Suggestions elsewhere that it is there as a physical barrier to protect the worksite.  So did it work?

Paul

Looking at the pictures, it seems a possibility that the engineering train was longer than was expected, or it had been moved, and that the passenger train then collided sidewise where the lines converge.
The damage looks limited perhaps suggesting that the driver of the EMU had time to brake, but not time to stop.
The EMU was leaving the station, so it must have arrived safely.

Under normal circumstances, track circuits on the line occupied by the freight should be interlocked with the platform starting signal and thus prevent this sort of accident.
During large scale engineering work, it is possible that the equipment was defective or disconnected, or even that it was BELIEVED to be defective when in fact it was working correctly.

I've been in and out of Waterloo half a dozen times recently and that train has been an almost permanent feature, at the current length.  The line it is on is within the possession until well beyond the point of impact.

It can be seen present, (either with the cab just at the platform end, or just beyond) in a number of online images and videos taken at different times, e.g. this from 8th Aug by Paul Bigland: http://PaulBigland.zenfolio.com/p671909530/e92a8b12d

Paul

Causing horrendous problems getting into Waterloo from suburban stations now. A colleague's journey from Surbiton this morning kicked her out at Raynes Park, and eventually told to get a bus to Wimbledon & the tube from there. Expected at 0930, she's still on her way.

But the engineering train shouldn't affect the signalling, should it? All track within the possession should be off limits, - I'm not sure whether that would involve screwdrivers and crocodile leads these days, but the effect should be the same. The issues concerns the boundary - all track entering the possession should appear "occupied" permanently, and be marked by sticky tape on the panel (or its soft equivalent on-screen).

Quite right AFAIK, but my post was only intended to explain to the earlier poster that that particular engineering train shouldn't have caught anyone out, because this morning it appears to be not significantly different, in either length or position, to any other day.

Paul

SWT are now advising not to travel on their network tomorrow as well as the rest of today:
https://www.southwesttrains.co.uk/disruptions/96AF36FED17149D9BF98F22568F054DC/

It will be interesting what the RAIB come up with as the underlying cause but the barrier train did exactly what it was supposed to do and probably prevented deaths or serious injuries on the ground in the work site. Good use of belt and braces against, what I can only presume, was a perceived risk of incursion - or, more likely, worker excursion.

From my very untrained eye, it looks to me that a set of points leading away from the platform from which the EMU was leaving had somehow been set so that the EMU went into the path of the engineering train
The points must have somehow been set between the time the EMU entered the platform and left it.

That looks like some evidence of their having been two issues (as was reported on SWT's website yesterday morning), perhaps the points giving access to P13 from the P14 approach track are a separate fault?  P13 is now showing NOGO on the opentraintimes map, and as you say P11 and P12 are seeing normal movements via yesterdays main incident site.

http://www.opentraintimes.com/maps/signalling/wat

Paul

To the best of my knowledge that has been the case all day, hence my earlier post. 

Looking at the track diagram, I suspect that if the "double slip (diamond) crossing", (which was the scene of yesterdays incident), is still defective, and it was secured in the 'normal' position for access into P11 and P12, then it would prevent use of P13.

Maybe one of our signalling gurus would like to offer a better opinion?

Paul

What ever happened to clipping, scotching and padlocking points in such circumstances?

I think that's what Paul meant by securing the points in the 'normal' position.

Yes - that's what I intended.   I'm sure the RAIB report will be interesting in due course.

Paul

SWT's excuses page has this:

That appears to say the the "unrelated" points failure yesterday was at the other side (though I can't see any trains that were affected), while this P11-13 issue has happened before. (Could that be what happened last week?) 

It gives the impression that work has been going on at the entrance to P13, though that seems a bit unlikely. More like sitting thinking and planning what to do later? Whether what needs mending at or near that crossing was the cause of the derailment, its result, or due to the rerailing operations is unclear, and may not be 100% clear to those on site.

It was obvious in an earlier post that I didn't know what technology was involved in signalling at Waterloo, more specifically in the interlocking which is what seems to have failed yesterday. The blurb about this upgrade programme said control was going to Basingstoke ROC, and even that this would be done starting last year. Yet the Sectional Appendix (June 2017) still says Wimbledon SC. That would still count as remote, I imagine, but what is it?

Waterloo is a 1990 relay interlocking controlled from Wimbledon SCC by duplicated Time Division Multiplex (TDM) remote control and indication systems.  The interlocking is divided into 'Main Line' and 'Windsor' sides and each half is controlled by separate panels at Wimbledon.

I am aware of the possible cause of the incident and rest assured the RAIB report will be very interesting reading.......

Yes Paul, although I don't know the exact reason, its quite likely that that is the cause as the only route to Platform No.13 passes through those points whereas all other (functional) platforms have alternative routes to/from them.

Do we know whether the derailment has caused any delays to the programme of works being undertaken?

I think if anything the civils work is progressing anyway.  Someone suggested in another forum any knock on effects will more likely be to the stage work on the signalling alterations.

I was looking at various webcam pics over the last few days and (at least to my eye) the platforms and track alterations seem on the home straight now; I think they are about to start the top surfaces of the 3 modified islands.

Paul

Yes, I've seen the way the panels are currently split, and it suggests an interesting medium term problem because the station and approaches effectively operate now with a split down the middle, with the main line (fast and slow) side covering P1-14, and the Windsor side panel covering the rest including P20-24, and the two signalling diagrams only have limited overlapping information.   

However, in the post Dec 2018 set up, the mainline side spreads out from P1 as far as P18 - with the main suburban apparently having P1-6, and the longer distance trains P7-18.   The Windsor side services will use the same split of approach tracks through Vauxhall, but their trains will only normally use P19-24. 

IIRC this sort of horizontal split runs all the way through to Clapham Junction - which as most people know doesn't really operate as a junction, more of a 'coming together'...  ;D

Sounds like they might need some significant changes to the signallers MMI...

Paul

Yes, the split continues through Clapham Junction to Earlsfield on the Mains and to Putney on the Windsors.  There is a seperate 'Yard Shunters Panel' at Clapham Junction that interfaces to both Wimbledon panels and to signal a train, for example, from the Mains, through the Yard, and back out to the Windsors requires some complex electrical controls and the co-operation of the three signallers!  I know this because I was Project Engineer for the design team back in 1989-1992 ;) :P

I would guess that when the reconfiguration of the control centre to VDUs comes on stream that it will be easier to implement a different split.

Platform 13 remains out of use again today so would appear they couldn't fix the problem overnight.

On the WNXX Forum it was mentioned that the derailed train (once rerailed) was stabled in Platform No.13 so perhaps its still there?

I wonder whether perhaps the points were set wrong when the inward working entered P13 some 20 minutes earlier.  This “trailing” of the points would not of course have caused a derailment, but would have damaged them and could have lead to the 0540 outward working derailing when passing over them. 

Partial/complete replacement of the moving bits of the points may be necessary, which they may have decided to postpone for now.

I believe it ran to Clapham Yard yesterday morning about 0815, 5Z50 was the train ID mentioned elsewhere. I think I saw it in P11 on the opentraintimes map beforehand but cannot be sure.

Paul

Not sure about this 'inward working to P13'?  The incident train was leaving P11, there have been various images online showing the majority of the train still in P11 after the event.  Another unit was trapped in P12 after the incident, and P13 was inaccessible, but IIRC empty at the time.

This Guardian article shows the aftermath, the barrier train being in the P10 approach:
https://www.theguardian.com/uk-news/2017/aug/16/rail-passengers-second-day-delays-waterloo-derailment

Paul 

from the SWT journey check

That's just as you'd expect, (as post #47) - hence P11 and P12 are in use, (trains in there as I write) which indicates that the affected crossing is presently out of action and must be locked in position for 'straight across' moves from P11 or P12 towards the Up Main Relief.

It doesn't mean the original incident concerned trains entering or leaving P13 at all.

Paul

Sorry, my mistake, I should have said P11 not P13.

...[later]  and RTT shows the train in P12 arrived after the train in P11, so that's my theory blown out of the water!

No problem.  We are back on the same song sheet again...   :)

Paul...

In that London Connections piece, it says:

I'd been thinking that was at least a likely reason, and also speculating why - probably that plan A for blocking that path (clipping) was impractical for just those bits of a double slip crossing, and plan B (temporarily changing the interlocking) might also be too difficult. The comments after that article suggest the same things, and add that testing any modified interlocking was a big job in itself.

If you look at the way the possession has changed each weekend, not just for the bank holiday, that means that whatever was done to that crossing needed to be undone and redone several times. That would have added to the difficulty of either plan.

During each weekend, the possession covers all four main lines out to Queenstown Road, leaving the passenger service only three tracks - fingers crossed nothing breaks there! It also means all possessed platforms are reached via a sizeable parking area for engineering trains. During the week, only the slow lines are taken over, meaning that platforms above 7 can only be reached via the West Crossings. But that's how trains in service get between the Main Fast Lines and their platforms. I'm guessing that was seen as the main danger point, though not one that can really be avoided.

From RAIB today (30th) 10h01
https://www.gov.uk/government/news/collision-near-london-waterloo-station-15-august-2017



News story
Collision near London Waterloo station, 15 August 2017
From:Rail Accident Investigation Branch Published:30 August 2017
Investigation into a collision near London Waterloo station, 15 August 2017.

Passenger train and wagons after the collision at Waterloo
Passenger train and wagons after the collision at Waterloo
At around 05:42 hrs on Tuesday 15 August 2017, a passenger train leaving London Waterloo station travelling at about 11 mph (18 km/h) collided with a stationary train of wagons. There were no injuries, but both trains were damaged and there was serious disruption to train services using the station.

The passenger train was the 05:40 hrs South West Trains service from Waterloo to Guildford and comprised 10 coaches, a combination of class 455 and class 456 electric units. The wagons were standing on a line adjacent to the intended route of the passenger train and had been provided as a safety barrier separating passenger services from an area where construction work was being undertaken as part of the Waterloo station upgrade project.

The collision occurred because a set of points were misaligned and directed the passenger train away from its intended route. The misalignment was a consequence of a temporary modification to the points control system which also caused the train driver and signaller to receive indications that the points were correctly aligned.

Our investigation will examine:

the circumstances leading to installation of the temporary control system modification
the safety measures provided while the temporary modification was in place
the checking and testing procedures applicable to the modification
any relevant underlying management factors.

Interesting to see this initial report ... removing the possibility of it being a SPAD or a signaller's error

To those of us who have undertaken risk assessments, it is always about one's position in the matrix bounded by likelihood - on the one hand - and consequences on the other, and what mitigation you employ in each part of that risk space. To my eyes, the barrier train did exactly what it was supposed to do in these circumstance.  It looks to me like it was the blunt, but all powerful, back-stop against a risk that was very unlikely, but with huge negative consequences if it happened.

I will be interested to see the RIAB look into the risk assessment programme.

Although the barrier train was more likely to prevent construction site RRV incursion onto the operational railway than operational rail vehicles incursion into the construction site.

Having been in many rolls which have involved signing off on work similar to this the report will be interesting to read to see if it was process or human failure or a combination of the two,

I agree you that an RRV excursion was always more likely, or a trolley for that matter.

I would also have  expected the point in question to be clipped and padlocked as well as relying on "A temporary modification to the points control system". Might burn the point motor out!

It is a double slip crossover, with four routes/positions.    IIRC from other discussions there are four separate points machines, and four possible positions of the crossover.   AIUI they cannot be partially clipped out of use because of the way the moving blades have to interact.   Clipping it in one position, say UMR to P12/13, would have taken P11 out of use. (or vice versa.)

Paul

That doesn't sound quite right. Surely the four routes only occur in two pairs, i.e. it has just two positions. If all the movements are linked, however many motors do the pushing, then you can't lock out one route without blocking the other route linked to it. If they all moved independently that might not be true - though detection might be an issue.

The point end approached by the passenger train that was incorrectly set is operated as a double slip pair.  The third end, controlled by the same point identity, was under the barrier train.  On a double slip each end is a pair that operate together from a single point operating mechanism.  I'll post up a drawing later.

In the meantime here is a video that shows the principle of operation.  Its not in the UK but the principles are the same: https://youtu.be/VpfJdm71u6g

Ah - so it is linkage (mechanical or not), but not what I said. After all, even if two routes can be set through that crossing at once, only one can be signalled - so there's no need to bother about their compatibility. It's more like linking the two ends of a crossover, and extending that to a third point end that can't usefully be moved independently.

And then trying to quickly alter it temporarily.

Did you manage to find a drawing?  I'm thinking the signalling panel engraved lines means something the way they are drawn, presumably they show the default straight route (i.e. P11 left <> DMR right) with everything set normal?

I had a quick glance at the hardware from a passing train and it looks like these are "clamp lock" operation, does that affect how the pairs are linked/operated?

Paul

Paul, still trying to find my drawings.  Might have to sketch it out myself......but have found a basic animation here: http://www.dccwiki.com/images/2/2f/CrossingDoubleSlip.gif

You are correct about the panel presentation and the fact the points concerned are Clamplock operated.

Thanks.   In the conversation about four positions, of course what we a looking at is analogous to a binary truth table, where Normal and Reverse are the equivalent of 0 and 1.  So two 'mechanisms' each with 2 conditions gives you four results?

N N
N R
R N
R R

...sort of thing?

Paul

Correct again (I'll make a signal engineer out of you yet ;) ).  In the incident the train was signalled through the points concerned NN, but they were actually set RN.  On clamplock operated double slips each end of a pair is individually driven by separate hydraulic rams but they are connected to the same single electro-hydraulic pump unit so 2 per double slip arrangement (there is a design version that has 2 separate electro-hydraulic pump units per pair - 4 in total for a set of double slips; but for reasons of economy in design, installation and ongoing maintenance, that configuration is rarely used).

Well, I'm glad to see that you two apparently understand it ...  :P :-[

Still doesn't answer why weren't the points clipped N N. If RN and RR would cause a collison

Its not for us to speculate why.  Its for RAIB to determine that.

I have tried to explain it in 'simple' non-technical terms but a double slip is probably the most complex set of points that exist so...... ::) :P

I have decided that I need to do you a drawing.  Watch this space  ;)

Because if clipped to straight through i.e. "NN" there would have been no access to P13?

Paul

Roger Ford's monthly preview of his next Modern Railways article again quotes an unnamed source who says that the points simply "should have been clipped" so I still await the RAIB detailed explanation.   

Paul

RAIB interim report.  I'm no expert - it is as expected all to do with stage works and testing, is very detailed and I'm hoping someone can follow it through and come up with a short precise.  Seems the point ends the train approached were physically 'mid position' as seen from the FFCCTV.

As was postulated in various forums at the time, the driver and signallers cannot be considered responsible in any way.

https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/669469/IR022017_171220_Waterloo.pdf

Paul

Primarily points 1524A and 1524B should have been clipped in their normal position (as per plan) but were not.

Secondarily, a test wire was left connected which shorted out the electrical detectors to 1524A&B and fooled the signalling system into believing the point blades were correctly set and locked normal.

If only one of the above had happened the accident would not have occurred - it required the combination of the two. Holes in the cheese.

Had unclipped point blades moved without the testing wire being present, detection would have been lost and the starting signal would not have cleared.

Along with the incident at Cardiff, these are gentle reminders to NR and it’s contractors that although the recent safety record is incredibly good, a major incident could still easily happen if safety procedures are allowed to slip.

10 out of 10 for that summary.  As a practicing signal engineer for nearly 50 years now, I am bitterly disappointed to have to read that the lessons learnt from the Clapham accident have not been followed through.  I know that the pressure to complete things to time is greater than ever, but when safety of the travelling public is concerned, we need to remember not to be frightend and to stand up when required and say NO.

With points 1524A and 1524B (and 1524C) all clipped normal, how would they achieve a route into P13/14?

Am I missing something obvious here?   Or did I misread something earlier that implied that 1524 and 1525 operated together?

Paul

Through Points 1525A which were free to move independently of points 1524A or B.  Points 1525B were also free to move but were located on a locked-out route.

Paul,  if you look at Figure 2 in the report you will see that it is 1525AB points that direct trains into/out of Platforms 12/13/14/15.  1524ABC points are only required normal for those movements.

Ok ta.   Goes back to that '4 state' truth table we discussed earlier doesn't it.

Paul

Correct.

One of the basic principles of signalling testing is that all Test Straps are individually numbered.  When you fit them you ‘count them all out and then count them all back in again’ to make sure one doesn’t get accidentally left in place.  If a ‘strap’ is required to be left into place after commissioning then it should be made permanent and designed/checked/approved within well documented procedures and standards.

Lets wait and see what the continuing RAIB investigation throws up.......

Yup. The number of wires landed on each terminal is in the plan.

Part of our [on my heritage railway] S+T sign-off, is confirming that each terminal has the correct wire count. Sounded anal to me the first time I had to do it, but this accident reinforces why it is so necessary.

Like SandTEngineer I am surprised that a test strap and points were left unclipped; whilst there is much ribbing of S & T by E & P one thing I do know from experience of working with S & T  testing teams is their attention to detail, being methodical and not being concerned if they tell the powers that be "we ain't finish testing yet"

Appears RAIB have produced their  Interim Report.

https://assets.publishing.service.gov.uk/media/5a3bc2b8ed915d61889d9b4e/IR022017_171220_Waterloo.pdf (https://assets.publishing.service.gov.uk/media/5a3bc2b8ed915d61889d9b4e/IR022017_171220_Waterloo.pdf)

Roger Ford has a summary in Febs Modern Railway.

Appears a Test Strap was left in place and the relevant points not clipped. He says similar to Cardiff incident in 2106 when a decommissioned set of points was not clipped.

Very worrying.

I agree, very worrying that they know an incident will occur in Cardiff in 88 years time.

I see low ticket sales for the service on that day. On the upside, a good one to short form!

...err.  We discussed the interim report in this thread from post #58 onwards..... ::) ;)

Second report just published https://www.gov.uk/raib-reports/report-19-2018-collision-at-london-waterloo

If anybody here struggles to understand any of it then please ask away and I'll try to answer/clarify.

Quite damming of NRs competence management processes (after all the recent cases on the preserved railway network).

In particular I was really concerned about this statement:


My involvement in Clapham was reasonably close and I always related this tale to anybody new that worked for me. At the time I was Project Engineer for the Waterloo Area Resignalling (WARS) at ML Engineering (Plymouth) Ltd. We had the contract for the Resignalling.  We were in the middle of installation in Clapham Relay Room (the new one for WARS located in Victoria Signalling Centre, yes, I know). New lineside cables had been installed terminating on racks, and the link terminals fitted with RED DOME NUTS (those in the S&T industry will know what they do). BR had stagework wiring installed on the Righthand side of the terminations whilst we were installing the new interlocking wiring on the Lefthand side, with the connecting link obviously not fitted. BR then used the new cables to connect the new trackside equipment being installed in stages to the old Clapham signalbox interlocking. So, one fateful morning, siting in my office having my day starting cup of tea, the telephone rang and the person at the other end said "Hey, have you heard there's been a big accident at Clapham Junction, three trains involved and lots of people killed and injured?". You obviously can guess that my heart certainly sank to the floor and thoughts started flashing through my mind, crikey hope our staff haven't done anything. Started looking at the news reports on the TV and you can then imagine how my feelings progressed during the day.

Well, I suppose my point is, it becomes one of those defining moments in your career that you NEVER FORGET. I just hope we don't end up having a repeat of all that to make people understand and learn it all over again.

Notwithstanding finding 152, I am surprised that not securing the points wasn't higher up in the list of errors.  If this had happened, then none of the subsequent events with test wiring etc. would have had any effect. The report appears to lean most heavily on the testers themselves - perhaps echoing shades of Clapham - but with a simple scotch and clip, they would have got away with it.