Lumo overspeed incident at Peterborough (17/04/2022)

[SuspectUsual]

As a layman, it seems very strange that the system should be designed so that the signal protecting the junction clears to green with junction indicator feather when the speed restriction on all the diverging routes is so big.

Now, I guess this wouldn't often be an issue, insofar as I suspect the vast majority of diverging trains will be calling at Peterborough and therefore there'd be no reason to accelerate excessively, but if the divergence is for the regulation of a non-stop train, it creates a risk.

What would be the downside of the approach controlled signal clearing to yellow for diverging routes?

 

[Wyrleybart]

As a layman, it seems very strange that the system should be designed so that the signal protecting the junction clears to green with junction indicator feather when the speed restriction on all the diverging routes is so big.

Now, I guess this wouldn't often be an issue, insofar as I suspect the vast majority of diverging trains will be calling at Peterborough and therefore there'd be no reason to accelerate excessively, but if the divergence is for the regulation of a non-stop train, it creates a risk.

What would be the downside of the approach controlled signal clearing to yellow for diverging routes?

I guess it does appear daft, but you have to separate driver route knowledge from signalling capabilities. The signalling system expects the guy sat at the front window to know the speeds of the track in front of him, so that the signalling offers clear unambiguous messages.

 

[dk1]

Are the regulating decisions something which should cause concern? From my layman's perspective, signallers should be free to route trains as they wish (within the bounds of route knowledge and traction clearance) without worrying that the driver is not going to know the route they sign.

Of course. Drivers route knowledge through stations and sidings along the route is not the signallers problem. If however a driver has not done a particular line or area of that route for a long period of time (or even ever) its quite acceptable to contact them to ask where the next signal would be. Better to ask than make a mistake.

 

[SuspectUsual]

I guess it does appear daft, but you have to separate driver route knowledge from signalling capabilities. The signalling system expects the guy sat at the front window to know the speeds of the track in front of him, so that the signalling offers clear unambiguous messages.

I understand that, but surely if a system can be designed so that it creates an additional protection against failure, without any other adverse consequences, it should be done? Its what we've done in my (different) industry for years

 

[43066]

I was always taught quite firmly that there's no such thing as an approach release signal, the correct term would be approach controlled and I certainly wouldn't expect an experienced driver to describe a signal as the former, especially on a Rules exam. If you teach drivers to expect that the signal will always 'step up', that's just a SPaD waiting to happen.

Approach controlled and approach release are two common terms for the same thing. Plenty of experienced drivers use the terms interchangeably, know exactly how the system works, and don’t have SPADs.

Nobody has suggested drivers be taught that the signal will always step up. Why do some trainers insist on assuming drivers will be unable to cope with the concept that a signal might step up (and we all know where they are) but you have to drive as if they won’t? Anybody who can’t cope with that shouldn’t be anywhere near the cab of a train.

As a layman, it seems very strange that the system should be designed so that the signal protecting the junction clears to green with junction indicator feather when the speed restriction on all the diverging routes is so big.

Now, I guess this wouldn't often be an issue, insofar as I suspect the vast majority of diverging trains will be calling at Peterborough and therefore there'd be no reason to accelerate excessively, but if the divergence is for the regulation of a non-stop train, it creates a risk.

What would be the downside of the approach controlled signal clearing to yellow for diverging routes?

It’s more restrictive than flashing aspects, as TPWS loops will be activated if you’re being checked down towards a red, whereas they won’t if you’re approaching flashing yellow sequence. We have flashing yellow sequences for junctions 110 down to a 30, and several 120/125 down to a 40. Any of which would likely cause loss of life if the driver hit them at linespeed. The system is very safe, but not not totally infallible.

Our TOC doesn't even acknowledge that approach controlled signals exist. You are simply told to drive assuming the red will stay red (and if that signal normally clears at a certain point that just happens to be chance, there is no reason for it).

Telling people untruths is not an accepted method of training in most industries!

 

[TurboMan]

Holding the junction signal at a restrictive aspect until the train approaches is what controls its speed through the junction

Except when it doesn't.

This is a standard arrangement all over the network. Drivers need to a. React correctly to the signal aspects theyre getting, and b. know the speed of the various turnouts on the junctions they sign.

Except when they don't.

We know that drivers, being human beings, are fallible and prone to error. So that even though (to paraphrase the quotes above) holding the signal at danger is what stops a conflicting move and collision, and drivers need to react correctly to the signal aspects they're getting, they very often don't, hence SPADs, and as a mitigation measure, TPWS TSS at stop signals. Yet there still seems to a reliance, as in this incident, on the driver being the sole mitigation against overspeeding over a junction, which could result in a derailment with potentially catastrophic consequences to a similar degree as with a collision caused by a SPAD.

The relevant standard (RIS-0775-CCS) allows for TPWS OSS to be fitted at diverging junctions like this, but there is an exemption for 'speed restrictions that apply to diverging junctions where risk assessment confirms that an approach-controlled signal aspect is sufficient to control the risk'. But the only way of controlling that risk without TPWS is by relying on the flawed human being in the seat to act on the signal aspects - that seems fundamentally flawed to me.

The basic principle of risk management is that that the human being is at the very bottom of the hierarchy, the very last step in controlling the risk, with engineering controls etc. being used in preference. I still don't think that concept is understood very well on the railway.


== Doublepost prevention - post automatically merged: 11 Jul 2023 ==

I understand that, but surely if a system can be designed so that it creates an additional protection against failure, without any other adverse consequences, it should be done? Its what we've done in my (different) industry for years

I quite agree, that's the point I was making in my post above.

 

[12LDA28C]

Of course. Drivers route knowledge through stations and sidings along the route is not the signallers problem. If however a driver has not done a particular line or area of that route for a long period of time (or even ever) its quite acceptable to contact them to ask where the next signal would be. Better to ask than make a mistake.

If a driver is driving over a line that they have never driven over then that would certainly not be acceptable from a risk assessment perspective and they'd find themselves in a spot of bother if they had an incident.

 

[43066]

I understand that, but surely if a system can be designed so that it creates an additional protection against failure, without any other adverse consequences, it should be done? Its what we've done in my (different) industry for years

We know that drivers, being human beings, are fallible and prone to error. So that even though (to paraphrase the quotes above) holding the signal at danger is what stops a conflicting move and collision, and drivers need to react correctly to the signal aspects they're getting, they very often don't, hence SPADs, and as a mitigation measure, TPWS TSS at stop signals.

The system is very safe. When did the last fatality arise as a result of the scenario in the accident report?

TPWS is by relying on the flawed human being in the seat to act on the signal aspects - that seems fundamentally flawed to me.

Perhaps, but it has still given us the safest railway in Europe. There is no business case for spending huge amounts to eliminate a tiny risk (which will eventually be naturally eliminated as ETCS is rolled out). Driver professionalism is still relied upon as a key safety component of the current system and will be for a very long time to come.

 

[4COR]

Even though it passed P486 without the flashing double yellow aspect being displayed, it could still have been shown a flashing single yellow aspect at P474 had it not been so close to the signal when the route from P468 was set. It is not necessary to inhibit a flashing sequence just because the driver wouldn't see the flashing double yellow.

Fair point - it's more clearly stated in the report that it's the occupation of the inhibiting track circuit (and it's proximity) that prevents the flashing yellow. It's not clear from the report (and probably not really relevant) how far past P486 that is (and possibly not so important in a simplified explanation?).

At 10:17:02 hrs, with train 1L06 still occupying the Up Fast line, train 1Y80 was travelling at around 32 mph (51 km/h) and slowing down as it approached and passed P486 signal displaying a single yellow aspect. Train 1Y80 then occupied track circuit T7726 which inhibited P474 signal from showing a single flashing yellow aspect (paragraph 40).

 

[Railsigns]

As a layman, it seems very strange that the system should be designed so that the signal protecting the junction clears to green with junction indicator feather when the speed restriction on all the diverging routes is so big.

What would be the downside of the approach controlled signal clearing to yellow for diverging routes?

- There would be no point, because the overall junction signalling arrangement is designed to control the train's speed for the diverging route irrespective of what aspect the junction signal steps up to.
- It could adversely affect performance.
- It would set a SPAD trap for the driver at the next signal (driver becomes accustomed to finding the next signal clear despite always passing the junction signal at yellow, then one day it isn't...)
- It would diminish the usefulness of the AWS indicator as a reminder to the driver because they would always get an AWS warning at the junction signal.
- It couldn't be applied consistently, because the next signal beyond the junction could be several miles away, and you wouldn't want to be running for miles on a yellow aspect.

 

[eldomtom2]

I understand that, but surely if a system can be designed so that it creates an additional protection against failure, without any other adverse consequences, it should be done? Its what we've done in my (different) industry for years

I think it's a bit late to resignal everything for speed signalling...

 

[bramling]

Approach controlled and approach release are two common terms for the same thing. Plenty of experienced drivers use the terms interchangeably, know exactly how the system works, and don’t have SPADs.

Exactly. The conventionally signalled lines on LU are absolutely full of signals with timing sections affecting how they will clear, and certainly some knowledgeable drivers would know exactly how these worked (to the point of counting 4.5 or 15 seconds after passing a particular point!). However they would never drive in such a way that they even came close to not being able to stop if the signal didn’t clear.

Yet plenty of people still did and do seem to find all this difficult to understand, almost like some people just never get how thermostats work. Hence there’s Piccadilly Line drivers who will wait for “two greens” when leaving Barons Court eastbound, even though this provides no guarantee at all that speed-controlled and multi-SPAD signal A655 is going to clear!

 

[Sly Old Fox]

If a driver is driving over a line that they have never driven over then that would certainly not be acceptable from a risk assessment perspective and they'd find themselves in a spot of bother if they had an incident.

That’s not necessarily true, I sign a few bits of track that I’ve never driven over or even travelled over. Nothing major, just goods loops and triangles and stuff.

 

[36270k]

A similar incident happened at Didcot in 1967. The MOT report is on railsarchive.

 

[12LDA28C]

That’s not necessarily true, I sign a few bits of track that I’ve never driven over or even travelled over. Nothing major, just goods loops and triangles and stuff.

Indeed, however whether or not you still 'sign' them having not driven them or had a refresher within the set timescale is open to debate.

 

[Snow1964]

The relevant standard (RIS-0775-CCS) allows for TPWS OSS to be fitted at diverging junctions like this, but there is an exemption for 'speed restrictions that apply to diverging junctions where risk assessment confirms that an approach-controlled signal aspect is sufficient to control the risk'. But the only way of controlling that risk without TPWS is by relying on the flawed human being in the seat to act on the signal aspects - that seems fundamentally flawed to me.

At the point when TPWS was added (don't know date but probably 10-15 years ago), the risk was clearly miscalculated, approach control was sufficient to slow the train to about 30mph, but not to keep it at a low enough speed by the time it reached the junction.

The distance from the signal (presumably before the over bridge to allow clear sighting), means there is about half a mile where unchecked acceleration is possible before a very low speed turnout (that is low speed as percentage of line speed). Clearly higher than average risk, but not treated as high enough to require TPWS OSS (the diverging junction speed warning).

The problem is not how well a driver knows a route, but if they make a mistake, the TPWS allows half mile of unchecked acceleration and nothing apart from driver realising late that train is now going to fast.

 

[Dieseldriver]

To my mind it seems prudent to have an additional warning board between the junction signal and the points with a suppressed AWS magnet for trains taking the primary route with it providing a warning horn when routed in the manner this train was. Should catch the attention of the Driver as it’s an additional reminder if they’d misread the junction and should give them the opportunity to bring the train back under control.
Obviously we’re talking about this very specific location as this is where the incident occurred, I have no doubt there are numerous other traps on the network (some even worse than this one) waiting to catch someone else out.

 

[skyhigh]

At the point when TPWS was added (don't know date but probably 10-15 years ago), the risk was clearly miscalculated, approach control was sufficient to slow the train to about 30mph, but not to keep it at a low enough speed by the time it reached the junction.

TPWS simply isn't designed to enforce a speed curve.

 

[dk1]

If a driver is driving over a line that they have never driven over then that would certainly not be acceptable from a risk assessment perspective and they'd find themselves in a spot of bother if they had an incident.

That’s not what I am saying. Im talking about certain yards/lines on the route that they may not have passed over. Absolutely fine to ask the signaller then if not sure. I’ve done it several times in my career and advise my trainees the same.

== Doublepost prevention - post automatically merged: 11 Jul 2023 ==

That’s not necessarily true, I sign a few bits of track that I’ve never driven over or even travelled over. Nothing major, just goods loops and triangles and stuff.

That’s what I am saying. For example I had never been through Ipswich Freightliner sidings but due to various issues got asked to pass through them with ECS. Signaller talked me through it & jobs a good’en. Still waiting to get diverted round the Bacon Factory Curve. I’d take that route in a heartbeat lol.

 

[357]

If a driver is driving over a line that they have never driven over then that would certainly not be acceptable from a risk assessment perspective and they'd find themselves in a spot of bother if they had an incident.

This is absolutely not correct.

 

[snowball]

There was an item of perhaps about 3-4 minutes on the PM programme on BBC Radio 4 at about 17:46 today.

 

[Lockwood]

I've got a probably quite stupid question, but...

What is it that determines if a junction gets a junction indicator like this or a theatre box?

Havant has a letter shown for route, rather than a position 1 indicator for Waterloo etc (or 4 for Brighton).

Why does this location have a signal with so many JIs and not an actual 1-6 shown to indicate platform? (and route knowledge for speeds from that)

 

[Dai Corner]

I've got a probably quite stupid question, but...

What is it that determines if a junction gets a junction indicator like this or a theatre box?

Havant has a letter shown for route, rather than a position 1 indicator for Waterloo etc (or 4 for Brighton).

Why does this location have a signal with so many JIs and not an actual 1-6 shown to indicate platform? (and route knowledge for speeds from that)

Another stupid question from a layman.

Why not have a a supplementary display showing the maximum speed when a feather for a diverging route is shown?

 

[Lockwood]

I'd guess that one would be differential speeds?

 

[dk1]

Another stupid question from a layman.

Why not have a a supplementary display showing the maximum speed when a feather for a diverging route is shown?

That should be known automatically to a driver that signs that basic route. I do not need to be reminded personally.

 

[MadMac]

I've got a probably quite stupid question, but...

What is it that determines if a junction gets a junction indicator like this or a theatre box?

Havant has a letter shown for route, rather than a position 1 indicator for Waterloo etc (or 4 for Brighton).

Why does this location have a signal with so many JIs and not an actual 1-6 shown to indicate platform? (and route knowledge for speeds from that)

There used to be a maximum approach speed for Theatre Indicators of 40 MPH - it also used to be that they had to display an indication for all routes including the “straight“ one, although I believe that’s no longer the case.

== Doublepost prevention - post automatically merged: 11 Jul 2023 ==

That should be known automatically to a driver that signs that basic route. I do not need to be reminded personally.

The other potential issue is that at some point, you get into the realms of “information overload“ where you give the driver so many things to process that they can easily miss something critical.

 

[ASharpe]

Are approach controlled signals only for stepping down the speed of the train? With a different signal further back protecting a block with all these signals in?

Or am I misinterpreting figure 8? It seems odd the delayed following train would be so close behind.

 

[dk1]

The other potential issue is that at some point, you get into the realms of “information overload“ where you give the driver so many things to process that they can easily miss something critical.

Not something that personally affects me or most drivers I know. It’s just what I do day in day out.

 

[Sheridan]

Are approach controlled signals only for stepping down the speed of the train? With a different signal further back protecting a block with all these signals in?

Or am I misinterpreting figure 8? It seems odd the delayed following train would be so close behind.

View attachment 138952

The screenshot you’ve posted only shows one train so I’m not quite sure what you’re asking. Each of these signals is capable of showing a red aspect therefore the space from one signal to another is a block and they function like any other signal in 4-aspect signalling. They are not there solely for the approach control, indeed they can all show green for the appropriate route.

 

[ComUtoR]

We got a way with it and that isn't a good enough response surely? ( agree on the interpretation point)

Sorry for the late reply. Out earning a crust.

There is always a lot of focus about what went wrong or trying to find where the blame is and how we can rip apart the system because its 'broken' in some obscure edge case way. Why not also look at what went right ? Yes, there is certainly a large element of luck but why didn't it tip over? How much was left in the bank so to speak ? What prevented a potential derailment and how can that be applied elsewhere.