Rulebook procedures following failure of onboard AWS/TPWS

I understood that a right-side failure was where the system failed into something more restrictive - like a "false positive" on a medical test - and a wrong-side failure was where it failed into something less restrictive - a "false negative".

So a right-side signal failure stops a train needlessly; a wrong-side signal failure fails to stop a train. A right-side AWS failure means it goes off inappropriately; a wrong-side AWS failure means it doesn't go off when it shouldn't.

The earliest semaphore signals used a downwards pointing semaphore for clear and a straight line for stop. This meant that snow piling up on top of it or the wires breaking would cause it to fail into the less restrictive aspect - a wrong-side failure. So they were redesigned (except for on the GWR) to make the signal for a clear line the semaphore pointing upwards, so that if it failed, it would at least fail into always saying stop - a right-side failure.

Kilopylae said:

I understood that a right-side failure was where the system failed into something more restrictive - like a "false positive" on a medical test - and a wrong-side failure was where it failed into something less restrictive - a "false negative".

So a right-side signal failure stops a train needlessly; a wrong-side signal failure fails to stop a train. A right-side AWS failure means it goes off inappropriately; a wrong-side AWS failure means it doesn't go off when it shouldn't.

Click to expand...

The AWS works by an inductor carried underneath the leading vehicles reacting to magnets in the 4ft. For normal direction of travel you have the permanent magnet followed by an electromagnet. The latter is denergised for Red, or Yellow aspects and this activates the AWS warning in the cab. Under a green aspect it is energised and that results in a bell in the cab. Thus if there is a failure of the electromagnet then a driver would receive a horn vice a bell when running under green signals. Thus its a right side failure ie it defaults to the safer option. Its possible for a driver to get caught out by an unexpected horn but they do have a few seconds to react before the equipment automatically brings the train to a stand.

Kilopylae said:

The earliest semaphore signals used a downwards pointing semaphore for clear and a straight line for stop. This meant that snow piling up on top of it or the wires breaking would cause it to fail into the less restrictive aspect - a wrong-side failure. So they were redesigned (except for on the GWR) to make the signal for a clear line the semaphore pointing upwards, so that if it failed, it would at least fail into always saying stop - a right-side failure.

Click to expand...

Correct and the wider principle of railway signalling has always been to design in right side failure into the system be it mechanical as youve described above or electrical and now software. However, the integrity of the system requires extensive testing when new and regular maintenance checks of every part of the system to make sure its functioning as intended, is within tolerance for technical equipment or in the case of cables that there is no low insulation resistance emerging that could cause a wrong side failure.

didcot_watcher said:

We perhaps need to remember Clapham. Driver stopped to report a faulty signal, that failed to protect his train.

Click to expand...

Unless I have misremembered, the collision at Clapham occurred while the driver was on the phone to the signaller reporting the fault. With trains running on three-minute headways, it would be a tall order to walk to the back of the train in three minutes, let alone do any meaningful protection.

Undaunted said:

Unless I have misremembered, the collision at Clapham occurred while the driver was on the phone to the signaller reporting the fault. With trains running on three-minute headways, it would be a tall order to walk to the back of the train in three minutes, let alone do any meaningful protection.

Click to expand...

It was pre GSMR and was a wrong side failure so correctly stopped to use the first available signal post telephone and tragedy struck whilst he was on the phone.

norbitonflyer said:

Not necessarily. If the line is clear and the AWS is working, but the signal is at fault, it could result in a bell but a non-green aspect - or no signal aspect at all.

Click to expand...

The AWS inductor (in relay-controlled areas) is driven off the aspect relay, or a repeat relay thereof (or should be), and also proves the signal alight. So, if correctly designed, a signal which hasn't cleared when it should won't result in bell for a non-green (including black) aspect, the two things are not independently controlled from the state of the line ahead.

didcot_watcher said:

We perhaps need to remember Clapham. Driver stopped to report a faulty signal, that failed to protect his train.

Click to expand...

Undaunted said:

Unless I have misremembered, the collision at Clapham occurred while the driver was on the phone to the signaller reporting the fault. With trains running on three-minute headways, it would be a tall order to walk to the back of the train in three minutes, let alone do any meaningful protection.

Click to expand...

I read that first comment as (correctly) blaming the signal, not the driver. As far as the driver knew, the signal was at red because a) he had seen it change to red as he was about to pass it, and b) his train was now occupying the section that it was supposed to protect.

Nicholas Lewis said:

It was pre GSMR and was a wrong side failure so correctly stopped to use the first available signal post telephone and tragedy struck whilst he was on the phone.

Click to expand...

Indeed. (The report notes that it might be suggested that perfect compliance with the rulebook would have required him to stop as soon as possible rather than continuing to the next signal, but says that insisting on that would be pedantic and unhelpful and the driver should not be criticised for his actions).

Signal Head said:

The AWS inductor (in relay-controlled areas) is driven off the aspect relay, or a repeat relay thereof (or should be), and also proves the signal alight. So, if correctly designed, a signal which hasn't cleared when it should won't result in bell for a non-green (including black) aspect, the two things are not independently controlled from the state of the line ahead.

Click to expand...

In guessing it's a delay in the relays that results in the known behaviour of the magnet still giving a warning for a couple of seconds after the green illuminates?

D365 said:

A failure of the onboard AWS/TPWS is presented to the driver by means of a ’simple’ fault indication.

Click to expand...

If the onboard AWS/TPWS system "knows" it is faulty (detects itself) it will display a Fault indication (sometimes combined with temporary isolation indication).

But "fault" detected by the driver could be a discrepancy between what they see and what the AWS does, eg if you pass magnets for a non-green signal and don't get a horn (this could be a failure of the trackside magnet, this could be a failure of the onboard AWS that the onboard system didn't detect in itself) - then the driver applies applies the Rule book "Code" response (normally stopping to report for a "wrongside" failure).

I think it's valid to consider whether a delay in reporting a potential trackside failure (i.e. NOT immediately stopping to report) is worth moving the reportinh train further from the "faulty" equipment/ signal to a safer location.

== Doublepost prevention - post automatically merged: 22 Jun 2026 ==

flitwickbeds said:

Can someone describe what "wrong side" and "right side" failures are, and how they differ/present themselves?

Click to expand...

In this instance:

If you get a Horn (caution) instead of a bell (clear) that would be a right side failure with a meaning expect cautionary aspect (even though true aspect is green)

If you get a bell instead of a horn, (or no horn), then the intended "expect cautionary aspect" indication isn't given - so a "wrongside failure"

A key learning may be that coming to a halt in the shortest distance possible after a suspected AWS fault may not be the best practice ....

Click to expand...

Seems it wasn't learnt at Clapham. Where as a commuter on that line I narrowly escaped death/serious injury.
Quoting from Wikipedia.

The Basingstoke train stopped at the next signal after the faulty signal, in accordance with the rule book. Railway historian Adrian Vaughan suggests this may not be the best way of handling faulty signals. If the Basingstoke train had carried on to the signal following the next signal, the crash would not have happened because the Bournemouth train would have stopped at the signal where the crash occurred. As of 1999, the rule book had not been changed.

Click to expand...

PeterBB said:

Seems it wasn't learnt at Clapham. Where as a commuter on that line I narrowly escaped death/serious injury.
Quoting from Wikipedia.

Click to expand...

Trouble with Clapham, it wasn't obvious to the driver of the train which stopped whether there was a train ahead or not. You cannot just carry on if you don't know the state of the line ahead

I assume that the train which stops has to contact the signaller - is there a reason why the signaller doesn't warn any train behind to slow to a crawl? Is it just that the headways are so short there isn't time?

bengley said:

The 'stop the train immediately and contact the signaller as soon as they become aware of a defect' refers to defects with on-train equipment, not on-track equipment. If a right side (AWS horn at a green signal) fault is received, this does not require the driver to stop immediately, but report at a convenient location.

Indeed, if the driver encountered a fault with the on train equipment, which is entirely possible, they must stop immediately as per above.

Click to expand...

I feel there’s some level of jumping the gun as far as this whole issue is concerned.

Clapham was nearly 40 years ago now, and I can’t readily think of any other incidents where a train stopping to report something has led to an incident occurring - though perhaps others will be able to point to examples?

We don’t yet have enough information on Friday’s incident to know whether this is a factor or not. It’s as yet unclear whether the 360 either
(1) saw a red aspect on WH154 at the last moment, braked but didn’t have enough room to stop (for example Watford)
Or
(2) disregarded a red aspect on WH154 and only braked upon seeing the train standing ahead (more akin to Ladbroke Grove)
And in either scenario we don’t know if there were any other factors such as a problem with the signalling equipment

The trains seem to have finished up some way south of WH154, but again we don’t yet know that extent to which the 810 might have been pushed along the track

I think it’s a little early to be considering rule changes until we have a clearer idea what actually went on. In any case, there will be situations where train defects cause brakes to apply with no input from the driver, in which case any decision on whether it’s a good idea to stop become academic.

In the case of Clapham, the driver that experienced the change from green to red didn’t really have a choice whether to stop or not. We know from hindsight that the signal was defective, but if a signal goes from a proceed aspect to stop it may well be because either the signaller has replaced it due to an emergency, or because something dangerous has occurred and the signalling system has reverted the signal back. There isn’t really the opportunity for the driver to be having a deliberation with themselves as to why the signal has reverted. In the overwhelming majority of situations it’s going to be best to stop.

bramling said:

I think it’s a little early to be considering rule changes until we have a clearer idea what actually went on.

Click to expand...

And I think we're getting into danger of being a little bit speculative about the Bedford Accident in general. We'll keep this thread going for now but we'd prefer it if this thread didn't go too far down the road of what might or might not have happened on Friday but stuck to the more general considerations of AWS/TPWS faults/failures and what happens or should happen.

The whole point of a driver stopping immediately in the event of a wrong side failure of the AWS or TPWS is so that their train is 'protecting' the fault - if I get a bell instead of a horn on the approach to a single yellow, for example, by stopping and reporting immediately no other train is going to reach that fault without being made aware of it by the signaller. It reduces the number of drivers who might experience that fault and therefore the amount of errors that can be made as a result.

Somewhere said:

Trouble with Clapham, it wasn't obvious to the driver of the train which stopped whether there was a train ahead or not. You cannot just carry on if you don't know the state of the line ahead

Click to expand...

I hoped it would be obvious, but any carrying on would be at a speed within visual stopping distance of any obstacle,
and not passing any junctions. (The station has a speed limit of 20mph btw)

PeterBB said:

I hoped it would be obvious, but any carrying on would be at a speed within visual stopping distance of any obstacle,
and not passing any junctions. (The station has a speed limit of 20mph btw)

Click to expand...

Which station? Clapham Junction?

dctraindriver said:

Which station? Clapham Junction?

Click to expand...

Yes. However, the junction part at Clapham is after the station for northbound trains,
and the tracks don't merge anyway, they run parallel.
The station is around 600 yds further on from the crash site.
Don't know how many more signals that would be, but at least one.
Vaughan's point seems reasonable.

PeterBB said:

Yes. However, the junction part at Clapham is after the station for northbound trains,
and the tracks don't merge anyway, they run parallel.
The station is around 600 yds further on from the crash site.
Don't know how many more signals that would be, but at least one.
Vaughan's point seems reasonable.

Click to expand...

I can’t talk for the Southern side but the SW side where the collision occurred isn’t 20mph. The main fast (where it happened) is 75 to 40 (15 into P7). Up slow is 60 mph into Clapham.

Plenty of speculation in today’s Metro newspaper, with the word “may” doing lots of heavy lifting … shame the mods can’t censor that !

AWS failures certainly used to be not unknown, due to the batteries for the electro-magnet part running down, and drivers became alert in days following to the ramp giving an incorrect warning.

I do wonder, from experience in the USA, whether the UK approach of minimal communication from trains to the centre being broadcast leads to a loss of awareness for others on the line. In the US radio is like air traffic control, everybody on frequency hears it, and can broadcast at any moment without needing to stop, calling out alerts to oncoming trains and such like. I've never heard of any USA incident in any way impacted by the ability to do this. Reporting a fault, and having the train behind hear there was an issue, would be done in seconds - in fact the dispatcher would ask the train behind to check.

Taunton said:

AWS failures certainly used to be not unknown, due to the batteries for the electro-magnet part running down, and drivers became alert in days following to the ramp giving an incorrect warning.

I do wonder, from experience in the USA, whether the UK approach of minimal communication from trains to the centre being broadcast leads to a loss of awareness for others on the line. In the US radio is like air traffic control, everybody on frequency hears it, and can broadcast at any moment without needing to stop, calling out alerts to oncoming trains and such like. I've never heard of any USA incident in any way impacted by the ability to do this. Reporting a fault, and having the train behind hear there was an issue, would be done in seconds - in fact the dispatcher would ask the train behind to check.

Click to expand...

US railways are a lot less safe than ours. This is old but probably still holds and shows they have more than 10x more deaths than us - https://pedestrianobservations.com/2011/06/02/comparative-rail-safety/

InkyScrolls said:

The whole point of a driver stopping immediately in the event of a wrong side failure of the AWS or TPWS is so that their train is 'protecting' the fault - if I get a bell instead of a horn on the approach to a single yellow, for example, by stopping and reporting immediately no other train is going to reach that fault without being made aware of it by the signaller. It reduces the number of drivers who might experience that fault and therefore the amount of errors that can be made as a result.

Click to expand...

Exactly. Imagine the rule does change and you carry on to the next "suitable location". Meanwhile the train behind not only encounters the same fault but also has a fault that brings the train to a stand anyway then you end up in the same situation.

PeterBB said:

I hoped it would be obvious, but any carrying on would be at a speed within visual stopping distance of any obstacle,
and not passing any junctions. (The station has a speed limit of 20mph btw)

Click to expand...

Another point is 40 years ago they did not have GSMR in the cab. As I understand it a single GSMR call can now get trains to a standstill?