Derailment at Dalwhinne

[whole diary]

[eightf48544]

This one seems to have slipped past.

https://www.bbc.co.uk/news/uk-scotland-highlands-islands-56711543

Seems they were doing somekind of tests with a short High Speed Train (HST^▸ (High Speed Train)).

Edit:VickiS - Clarifying Acronym

[bradshaw]

This is a thread on Twitter about it.

https://twitter.com/timonfrancis/status/1380936790715084801?s=21

[grahame]

From The Scotsman

RAIB^▸ (Rail Accident Investigation Branch) interim report - double line section, train signalled straight ahead but points set to 15 m.p.h. crossover. Going 30 m.p.h. ... could have been much faster; thank goodness it wasn't.  Next step - finding out why the points were wrong.   Suggestion also that signalman's systems showed train routed and signalled straight through.

[Oxonhutch]

Suggestion also that signalman's systems showed train routed and signalled straight through.

As an amateur S+T, that would be very disturbing - a wrong side failure.

[stuving]

Suggestion also that signalman's systems showed train routed and signalled straight through.

As an amateur S+T, that would be very disturbing - a wrong side failure.

it's a bit more complicated than that. If you look at the RAIB^▸ (Rail Accident Investigation Branch) words, the only valid route for this train in the Up direction was the Up line - the crossover was only available in the Down direction. So how is that directional protection implemented in a basically mechanical box (though this crossover is motor driven)?

The signal on the approach to the crossover was displaying a proceed indication for the route along the up line and there is no signalled route from there, over the crossover, to the down line. Both sets of points forming the crossover were detected as being in the correct position for the up-line route by the signalling system and were indicated as such to the signaller, even though the points at the north end of the crossover (the end nearest Dalwhinnie station) were set in a position to divert the train onto the crossover.

[grahame]

Suggestion also that signalman's systems showed train routed and signalled straight through.

As an amateur S+T, that would be very disturbing - a wrong side failure.

it's a bit more complicated than that. If you look at the RAIB^▸ (Rail Accident Investigation Branch) words, the only valid route for this train in the Up direction was the Up line - the crossover was only available in the Down direction.

Agreed on the complexity.  Still a very, very disturbing.  I would hope that the full / next layer of the investigation will get to the bottom of what happened; usually these investigations do that.  And from that, lessons learned and from those lessons, changes / updates so that our railways become even safer.

[Oxonhutch]

it's a bit more complicated than that. If you look at the RAIB^▸ (Rail Accident Investigation Branch) words, the only valid route for this train in the Up direction was the Up line - the crossover was only available in the Down direction. So how is that directional protection implemented in a basically mechanical box (though this crossover is motor driven)?

I agree that it will be complicated, and I made my 'wrong-side failure' comment before finding and reading the govt news report;

Wherein it writes:

Why the signalling system did not detect that the points were in an incorrect position thereby allowing the signal to be cleared for the movement along the up line

That is a classic wrong-sider when the interlocking fails to detect such a configuration. I will be interested too see the full report with diagrams, interlocking tables, photos and a detailed explanation of events that led up to this incident.

[eightf48544]

Interesting part of the RAIB^▸ (Rail Accident Investigation Branch) investigation.

"How the points were able to move as the train passed over"

Shades of West Ealing and Potters Bar.

[grahame]

From the BBC^» (British Broadcasting Corporation - home page) - https://www.bbc.co.uk/news/articles/cgl044x0m3eo

A train derailed in the Highlands after it was wrongly diverted from one line to another.

The train was being used for tests at station platforms when its three rear bogies came off the line at Dalwhinnie, at about 03:00 on 10 April last year.

The five people on board were not injured.

In a new report, the Rail Accident Investigating Branch (RAIB^▸ (Rail Accident Investigation Branch)) said the accident was caused by a wiring error that resulted in a failure of a signalling system.

Investigators said it was fortunate the incident had not been more serious.

Andrew Hall, chief inspector of rail accidents, said the signalling system did not detect that some points on the railway line were in an unsafe position.

He said: “Some of the causes of the accident at Dalwhinnie bear an alarming similarity to those found in the multi-fatal accident at Clapham Junction in 1988, and the more recent collision at Waterloo in 2017 which caused huge disruption on routes into London."

[stuving]

I'm a bit surprised by the lack of emphasis on this one of the identified causes:

Out-of-correspondence testing
136 The maintenance tester did not complete the specified out-of-correspondence testing following installation of the reserviced point machine.

[Out-of-correspondence testing, for this crossover, involves (at least) manually placing the points in all four positions, and the control lever in each position, and testing which of those eights states result in correct detection. It is the final back-up to every other check during the preparation and work.]

The corresponding recommendation is:

5 The intent of this recommendation is to reduce the likelihood of essential signal maintenance testing tasks being overlooked and not completed.

Network Rail should review its arrangements for recording progress when carrying out testing defined in its signal maintenance testing handbook. This should take into account environmental and other challenges relevant to the workplace and make enhancements that ensure practical contemporaneous recording of:
• the completion of each test step
• relevant test results, measurements, and findings (paragraph 177c.ii
and 177d).

However, I suspect this one may be more important:

4 The intent of this recommendation is to ensure that, when signalling maintenance teams make engineering changes to the signalling infrastructure, the requirement for the maintenance tester to be independent of the installers is effective in assuring the integrity of the signalling system.

Network Rail should review how it can best achieve the required level of independence between the installation and testing roles when pre-planned renewal work is carried out under the processes described in its signal maintenance testing handbook. This should take into account how people undertaking these roles work currently. It should make enhancements so that practical working arrangements are defined.

This recommendation may be relevant to other types of signalling work undertaken under arrangements described in Network Rail’s signal maintenance test handbook (paragraphs 177b.i, 177b.ii, 177c.i, 177c.iii and 178).

The issue is how you get two people working together in a small team to genuinely check each others' work, and challenge anything that seems to be missing or wrong. It's similar to the long-standing issue of making voice communications include formalities like "repeat back" - it feels awkward among friends.

[GBM]

I'm a bit surprised by the lack of emphasis on this one of the identified causes:

However, I suspect this one may be more important:

4 The intent of this recommendation is to ensure that, when signalling maintenance teams make engineering changes to the signalling infrastructure, the requirement for the maintenance tester to be independent of the installers is effective in assuring the integrity of the signalling system.

Network Rail should review how it can best achieve the required level of independence between the installation and testing roles when pre-planned renewal work is carried out under the processes described in its signal maintenance testing handbook.
The issue is how you get two people working together in a small team to genuinely check each others' work, and challenge anything that seems to be missing or wrong. It's similar to the long-standing issue of making voice communications include formalities like "repeat back" - it feels awkward among friends.

Shudder. Smacks of privatisation, and that to me is not good.
A race to the bottom with private companies undercutting each other to offer a cheaper service

[stuving]

One of the previous reports referred to in this RAIB^▸ (Rail Accident Investigation Branch) report was report 04/2010, "Incident at Greenhill Upper Junction, 22 March 2009". That incident was strikingly similar to the one at Dalwhinnie. In both, a replacement point motor was installed in a set with collective detection, but was configured with internal links to suit it for use alone or in a different position in such a set.

Both times there were shortcomings in the preparation and record keeping, and most importantly both times the out-of-correspondence testing was botched. At Greenhill the team did at least do the full set of tests (as far as the incomplete records show) - but the incorrect behaviour of the relay and panel indicator was not seen.

In the Greenhill case the four-man team were all from one local contractor, rather than all from Network Rail as at Dalwhinnie.