Wrong-side signalling failure at Wingfield

[John Webb]

News item from the RAIB:

On the morning of 26 October 2022, a signal at Wingfield in Derbyshire (which had been disconnected and reconnected the previous night as part of planned track maintenance work) was returned to service with a fault. This fault caused a wrong side failure, with the signal’s red and yellow aspects being displayed incorrectly.

The fault initially caused a train to pass the signal at danger, which stopped in the track section beyond the signal. It later resulted in a second train being signalled at caution into the track section where the first train was still stopped. The second train was running at slow speed and being driven at caution, and came to a stop 75 metres from the first train. No damage or injury was caused by the incident.

RAIB plan to issue a Safety Digest when their enquiries are completed.

Sorry - the link to the RAIB item: https://www.gov.uk/government/news/wrong-side-signalling-failure-at-wingfield

 

[D365]

Not a pleasant read…

 

[baz962]

Was just discussing this at work. Nearly much worse as the next train behind was nearly the third train in the section. Some very angry drivers this morning.

 

[43066]

I wondered when this would see the light of day!

EDIT: I’m not at all sure the second train was being driven “at caution”, AIUI it was being slowed down due to encountering a cautionary single yellow aspect in a three aspect signal section when the driver spotted tail lights... Not the same thing.

It’ll be interesting to see if that is clarified in the safety digest.

 

[Flying Snail]

A lucky escape by the sounds of it, that could have been very nasty indeed.

 

[skyhigh]

Was just discussing this at work. Nearly much worse as the next train behind was nearly the third train in the section. Some very angry drivers this morning.

I honestly can't understand how this was allowed to happen. Surely testing should ensure that the aspect displayed is correct, at the very least?

 

[Purple Train]

Cardiff East Junction irregularity, Waterloo collision, Dalwhinnie derailment, now this.
We've been lucky so far with the consequences of these incidents involving wrong-side failures. Will it take another Clapham to stop that?

 

[Flying Snail]

Was just discussing this at work. Nearly much worse as the next train behind was nearly the third train in the section. Some very angry drivers this morning.

I don't blame them. Not easy to sit at the front of a train at high speed if there is even a slight doubt in the mind about the safety of the signalling.

 

[ComUtoR]

Cardiff East Junction irregularity, Waterloo collision, Dalwhinnie derailment, now this.
We've been lucky so far with the consequences of these incidents involving wrong-side failures. Will it take another Clapham to stop that?

They will never stop. Failures happen all the time, some are significant and some just minor. What happens, and does, is that each time we learn from each incident and put mitigations in place and change procedures that don't work. The RAIB will do their investigation and make recommendations.

 

[philthetube]

This does not sound like a failure in the normal sense, this was an error by someone, a failure should always fail safe.

 

[edwin_m]

This does not sound like a failure in the normal sense, this was an error by someone, a failure should always fail safe.

Many failures are caused by human error - it's still the correct way to describe them. When a failure doesn't fail safe it's described as "wrong side failure", as given in the OP.

 

[QueensCurve]

They will never stop. Failures happen all the time, some are significant and some just minor. What happens, and does, is that each time we learn from each incident and put mitigations in place and change procedures that don't work. The RAIB will do their investigation and make recommendations.

Yes, failures are routine, and the system is designed to fail to safety.

What is worrying is the proliferation of wrong side, ie to danger, failures. There are too many of them and, unless some serious learning occurs to prevent them, there will be a serious event. The classic example of course being the Clapham Junction double collition in 1988.

 

[bramling]

They will never stop. Failures happen all the time, some are significant and some just minor. What happens, and does, is that each time we learn from each incident and put mitigations in place and change procedures that don't work. The RAIB will do their investigation and make recommendations.

This is a little different, as this seems to have been a wrong-side failure, and there should be processes in place to prevent things reaching that point, and furthermore those processes should be robust enough to make them unbreakable.

 

[ComUtoR]

What is worrying is the proliferation of wrong side, ie to danger, failures. There are too many of them and, unless some serious learning occurs to prevent them, there will be a serious event. The classic example of course being the Clapham Junction double collition in 1988.

This is a little different, as this seems to have been a wrong-side failure, and there should be processes in place to prevent things reaching that point, and furthermore those processes should be robust enough to make them unbreakable.

I've seen a fair few on our patch this year.

 

[bramling]

I've seen a fair few on our patch this year.

Would be interested if you feel able to post more details (but fully understand if you don’t).

 

[philthetube]

They will never stop. Failures happen all the time, some are significant and some just minor. What happens, and does, is that each time we learn from each incident and put mitigations in place and change procedures that don't work. The RAIB will do their investigation and make recommendations.

I've seen a fair few on our patch this year.

It sound as if this learning is not happening then.

 

[Nicholas Lewis]

If the SMTH handbook (Signal Maintenance Testing Handbook) is followed this shouldn't happen. This is why its a RAIB digest nothing significantly new to learn.

When it is clear that the safety learning from an event has been identified by a previous investigation or relates to compliance with existing rules, we may choose to publish a safety digest, rather than carry out a full investigation.

 

[ComUtoR]

It sound as if this learning is not happening then.

Each investigation will be different. Each incident will be different. Whilst there should be some industry level learning, the human error factors are hard to remove. Miss-communication, out of date documents, lack of training, fatigue, etc. are all potential factors. The learning part is easy. Implementing change is very different.

 

[Falcon1200]

The RAIB news story raises more questions than it answers! For example;

What aspect should the signal have been showing?
What aspect was it actually showing - Was it Red and Yellow at the same time?
Why was a second train cautioned towards the first train? Regardless of the signal issue, were the track circuits operating incorrectly also?

 

[MarkyT]

Sounds like lamp units may have been reconnected the wrong way round in some part of the circuitry if a yellow was being displayed when the signal should have been showing red. This type of error is practically impossible to protect against technically. Procedures require a proper check after reconnection that the actual observed output state correctly matches what the control system thinks it should be. Worryingly, some steps must have been missed in this case.

== Doublepost prevention - post automatically merged: 11 Nov 2022 ==

The RAIB news story raises more questions than it answers! For example;

What aspect should the signal have been showing?
What aspect was it actually showing - Was it Red and Yellow at the same time?
Why was a second train cautioned towards the first train? Regardless of the signal issue, were the track circuits operating incorrectly also?

If some weird combined aspect was displayed, the driver should treat that as a red, just like a signal encountered when completely extinguished.
Data logging records should reveal what the interlocking states throughout the incident were. I think the phrase 'signalled at caution' means a cautionary aspect (i.e. yellow) was displayed into an occupied section, which should be prevented by the interlocking normally of course.

 

[43066]

This is a little different, as this seems to have been a wrong-side failure, and there should be processes in place to prevent things reaching that point, and furthermore those processes should be robust enough to make them unbreakable.

This was genuinely one of the worst I’ve ever heard about.

 

[HSTEd]

Did the first train stop because of a TPWS triggering?

Ultimately the only way to avoid this sort of thing is to remove the signal heads from the equation.

The fewer components there are in the signalling chain to go wrong the better.

 

[43066]

Did the first train stop because of a TPWS triggering?

No.

Ultimately the only way to avoid this sort of thing is to remove the signal heads from the equation.

The fewer components there are in the signalling chain to go wrong the better.

We have the next fifty years or so to get through first…

 

[pdeaves]

Ultimately the only way to avoid this sort of thing is to remove the signal heads from the equation.

The fewer components there are in the signalling chain to go wrong the better.

The alternative could well have more 'things' to go/get wrong.

 

[MarkyT]

Did the first train stop because of a TPWS triggering?

Assuming the signal concerned is TPWS equipped, and depending on what and where the rewiring error actually was, then it's possible OSS and TSS loops were not activated for the incorrect yellow vice red aspect. Protection systems often derive their input triggers using current or voltage sensing at the same cable terminals that are connected to the signal lamps.

 

[Wilts Wanderer]

The alternative could well have more 'things' to go/get wrong.

Absolutely this - and the potential errors may be hidden in specialist software that only a few people in the world know how to interrogate.

 

[HSTEd]

Absolutely this - and the potential errors may be hidden in specialist software that only a few people in the world know how to interrogate.

Those errors will already exist given that essentially all modern interlockings are controlled by software anyway.

== Doublepost prevention - post automatically merged: 11 Nov 2022 ==

No.

If not.... why did the first train stop rather than continuing forward? It's not clear from the article linked.

 

[Bald Rick]

As per the RAIB, it is my understanding that the first train had the signal revert to danger on approach, hence they stopped. The second train was then cautioned.

 

[MarkyT]

If not.... why did the first train stop rather than continuing forward? It's not clear from the article linked.

It is reported the first train went past a red. This must have been authorised verbally by the signaller who, confusingly, may have seen the signal indicating proceed on their control screen/panel, if signals are separately indicated in the area concerned, or, if the signal is not separately indicated the signaller might have seen no reason from track circuit occupancy why the signal shouldn't clear and might have assumed a normal right side local trackside fault was somehow holding the signal at red. It's possible the signaller elected to hold the next signal at danger while they tried to figure out what was going on, or the signal couldn't clear for some other reason hence the first train stopped again. Once the first train passed the incident signal, it would have changed automatically to (false) yellow instead of staying at red, as it would be expected to, with the signaller still none the wiser. When the second train arrived, a yellow caution aspect was still mistakenly displayed instead of the correct red. Luckily, after entering the occupied section, the second train must have been going sufficiently slowly to be able to stop within the driver's forward visibility at the point the obstruction ahead was spotted.

 

[edwin_m]

Seems a rather similar scenario to Clapham, where the driver noted an irregular aspect and stopped at the next signal to report it. The first signal then gave a false proceed aspect to the following train.