Overspeeding incident at Grantham South Junction (RAIB investigation) -- 25/2/25

[HSTEd]

We do appear to have reached the practical limits of TPWS as a solution.

Perhaps you could do something with ETCS L1LS, which could impose a speed restriction of arbitrary length on the diverging route using a switchable balise.

But even if we assume all ECML (or all trains generally) trains will end up with ETCS gear, switching to L1LS would probably still delay the rollout of the L2 system. Although the Swiss did replace their legacy train protection systems quite rapidly.

 

[Tim_UK]

There comes a point where you spend more on bodging than just doing it properly. And if ETCS is the way to do it properly, then crack on. ETCS will evolve, and people will get better at designing and using it. And suppliers can find ways to ETCS upgrade existing kit.

There will be set backs, problems, the unexpected. Hassle from the upgrade. But once alll the busy routes are done, nobody will look back.

Outside of rail, I've seen so many projects where people hang onto something old, because `it works`. We can just bodge one little bit more. The cost of every bodge just gets higher. Every excuse in the book about why not to use the new stuff. But 90% of the time, once you commit and start you realise it's challenging but possible. The further you go, the easy it gets. The world moves on.

 

[Falcon1200]

What about a (possibly!) simpler solution, akin to the additional SPAD signals installed at a few locations? At a suitable point, ie braking distance, before the diversion, an indicator lit only when the diverging route is set, flashing and displaying the speed through the points.

 

[zwk500]

We do appear to have reached the practical limits of TPWS as a solution.

Perhaps you could do something with ETCS L1LS, which could impose a speed restriction of arbitrary length on the diverging route using a switchable balise.

Although I think there might be specific cases where L1 LS could be a solution, AIUI the ETCS concept is that you are either in an ETCS area or not, and therefore it's not quite so simple to fit ETCS at individual signals as it would be TPWS/AWS. There's almost certainly a way around it, but then it gets into 'why not just go ETCS L2 Overlay?'.

But even if we assume all ECML (or all trains generally) trains will end up with ETCS gear, switching to L1LS would probably still delay the rollout of the L2 system. Although the Swiss did replace their legacy train protection systems quite rapidly.

Tbf fitting of trains has to happen for any ETCS project, so isn't likely to be a major sticking point. The conversion of the infrastructure, however, is another matter. Didn't the Swiss spend quite a lot of time aligning their legacy systems (similar to the Belgians) so that the switchover could be done more smoothly? (Worth noting the first deployments of ETCS L2 in Switzerland were not without their problems, and as ETCS has matured in Switzerland the rollout has gone smoother).

 

[Nicholas Lewis]

There comes a point where you spend more on bodging than just doing it properly. And if ETCS is the way to do it properly, then crack on. ETCS will evolve, and people will get better at designing and using it. And suppliers can find ways to ETCS upgrade existing kit.

There will be set backs, problems, the unexpected. Hassle from the upgrade. But once alll the busy routes are done, nobody will look back.

Outside of rail, I've seen so many projects where people hang onto something old, because `it works`. We can just bodge one little bit more. The cost of every bodge just gets higher. Every excuse in the book about why not to use the new stuff. But 90% of the time, once you commit and start you realise it's challenging but possible. The further you go, the easy it gets. The world moves on.

We need to adopt some of the European countries approach whereby the operators get told this line will only be available to ETCS equipped stock from this date and thats the end of it not the nonsense we have we 10's millions being spent to motive power that will rarely use it but thats holding back implementation. The other thing that must happen on future extensions is its signals away no more expenditure on new signal structures which have grown ever larger and expensive to install.

 

[HSTEd]

Although I think there might be specific cases where L1 LS could be a solution, AIUI the ETCS concept is that you are either in an ETCS area or not, and therefore it's not quite so simple to fit ETCS at individual signals as it would be TPWS/AWS. There's almost certainly a way around it, but then it gets into 'why not just go ETCS L2 Overlay?'.

ETCS L2 Overlay has substantial costs beyond L1 LS though. Continuous supervision of speed and the like is hardly cheap, the LS specification does not require universal installation to my understanding, merely that no other systems are in use.
An L1 LS conversion ("merely") requires a switchable balise in place of each switchable AWS or TPWS unit, and a fixed balise in place of each fixed AWS magnet.

Indeed, the quality of surveying required for the Level 2 installation will be a substantial cost.

Tbf fitting of trains has to happen for any ETCS project, so isn't likely to be a major sticking point. The conversion of the infrastructure, however, is another matter. Didn't the Swiss spend quite a lot of time aligning their legacy systems (similar to the Belgians) so that the switchover could be done more smoothly? (Worth noting the first deployments of ETCS L2 in Switzerland were not without their problems, and as ETCS has matured in Switzerland the rollout has gone smoother).

They apparently did, "EuroSignum" and "EuroZUB", although I believe this was more about allowing a transparent conversion with work between trains.
There are about ~12000 TWPS fitted signals in the UK, I don't really have a figure on AWS but lets just guess 30000.
42000 replacements to eliminate TPWS and AWS - I think it is probably a more tractable solution than a large scale L2 programme in the near term (although obviously lower reward!)/.

Two staff could install the drop in self powered ZUB/Signum solution in two hours, supposedly. So, supposedly, about ~170,000 staff-hours trackside for such a programme. One could theorise about some sort of "rolling closure" programme moving along a line.

 

[zwk500]

We need to adopt some of the European countries approach whereby the operators get told this line will only be available to ETCS equipped stock from this date and thats the end of it not the nonsense we have we 10's millions being spent to motive power that will rarely use it but thats holding back implementation.

Which other railways have done this, and how has it gone for them?
NR will be loathe to kick freight off the network because once its gone, its goong to be very hard to get it back. Whereas by fitting a loco early, it will only increase the value of later ETCS fitment of routes.

The other thing that must happen on future extensions is its signals away no more expenditure on new signal structures which have grown ever larger and expensive to install.

There's also been considerable development of lightweight signal structures and some posts on social media from industry figures I've seen bemoaning the overblown safety have been rather disingenuous. You'd think there'd never been a semaphore offset bracket, let alone the Rugby Bedsteas, to hear some people talk of it.

ETCS L2 Overlay has substantial costs beyond L1 LS though.

substantual benefits as we though.

Continuous supervision of speed and the like is hardly cheap, the LS specification does not require universal installation to my understanding, merely that no other systems are in use.

Ah I didn't realise LS could be an add on.

Indeed, the quality of surveying required for the Level 2 installation will be a substantial cost.

Indeed

They apparently did, "EuroSignum and EuroZUB", although I believe this was more about allowing a transparent conversion with work between trains.
There are about ~12000 TWPS fitted signals in the UK, I don't really have a figure on AWS but lets just guess 30000.
42000 replacements to eliminate TPWS and AWS - I think it is probably a more tractable solution than a large scale L2 programme in the near term (although obviously lower reward!)/.

Two staff could install the drop in self powered ZUB/Signum solution in two hours, supposedly. So about ~170,000 staf-hours trackside for such a programme. One could theorise about some sort of "rolling closure" programme moving along a liline.

The issue presumably is whether the signalling and protection systems can cope with essentially 'random' signals having ETCS and everything else having TPWS. If LS can be fitted on a signal by signal basis then as a stopgap at critical overspeed risks I think it'd be worth looking into.

 

[HSTEd]

The issue presumably is whether the signalling and protection systems can cope with essentially 'random' signals having ETCS and everything else having TPWS. If LS can be fitted on a signal by signal basis then as a stopgap at critical overspeed risks I think it'd be worth looking into.

Yeah, I'd expect you'd have to completely replace the TPWS and AWS on the line with ETCS L1LS to make it work properly.

But I suppose the Swiss example raises questions about whether that might not be the best choice anyway.
If ETCS L1LS becomes the train protection system on the railway, the AWS and TPWS specific training burden and spares burden can be jettisoned.

"Only" around 12000 TPWS fitted signals at the end of the day.

 

[Nicholas Lewis]

Which other railways have done this, and how has it gone for them?

Czeckia have implemented it on a number of main line routes already and freight operators have to provide compatible power. Same for Luxembourg and some stretches of freight lines in holland, and the alp tunnels in Switzerland. The advantage here is freight operators have leased either Traxx or Vectron locos which have ETCS L2 as standard. Some TGV routes are now converted as well.

NR will be loathe to kick freight off the network because once its gone, its goong to be very hard to get it back. Whereas by fitting a loco early, it will only increase the value of later ETCS fitment of routes.

You dont kick it off you tell it you've got x years to get compliant locos. So its right that a standard 66 fitment is progressed. I believe the 99s are coming with ETCS.

There's also been considerable development of lightweight signal structures and some posts on social media from industry figures I've seen bemoaning the overblown safety have been rather disingenuous. You'd think there'd never been a semaphore offset bracket, let alone the Rugby Bedsteas, to hear some people talk of it.

Yup for standard one signal posts they've certainly made improvements that avoid working at height now but anywhere a gantry is involved the structures are humungous compared to what BR used to use. That approach is necessary now as us humans have got a bit heavier humans and the fact that the structures have to be able to support two maybe three staff simultaneously. Then there is large foundations that are then required to support such a structure. Whatever the efforts to improve every signal needs survey foundations installing, structures erecting and cabling up and much more importantly sighting which in itself is very involved task. So all this can be saved in future if resignalling schemes are signals away as well as better optimising the block sections.

 

[edwin_m]

Some recent signal gantries have no built-in access arrangements and are therefore much lighter. With LED signals, maintenance access is rare and if it becomes necessary they will have to take a possession and bring in a RRV.

 

[Railperf]

Controversial opinion incoming...While the reference to higher performing trains is a valid one, there are other power settings we can select other than 'Max'

When the Class 800 fleet was new, a lot of drivers at my TOC correctly stated that the speeds out of London that were not attainable by the former fleet would easily be reached or exceeded with the Class 80x if not careful. I couldn't have agreed more, so my solution was to select approx 30-40% power and just let it run until posted line speeds exceed 95mph. Doesn't bust any speed limits or lose time.

When it comes to departing stations, the power stays in the lowest setting until I can be sure the train is clear of the platform. That first notch of power is still quicker than what the previous fleet could achieve from a stand, so there's no detriment to the timetable and it's smoother for all involved. The over-speeding incidents that have recently occurred on our network seem to have come off the back of drivers going straight to full power in situations that really don't warrant it.

I had an Avanti driver doing just that driving the 1621 Birmingham to Euston - on a service that needs full acceleration and reasonably confident braking to minimise time loss on a schedule it cannot meet reliably. Needless to say we lost more than the usual time on each section. Passengers were not impressed.

I recall someone at LNER telling me that power notch 4 was to be used when departing stations. And a pal who is a driver instructor on 80x at another TOC says driving policy is that full power be used as soon as the platform monitoring cameras switch off around 5mph.

 

[TreacleMiller]

Controversial opinion incoming...While the reference to higher performing trains is a valid one, there are other power settings we can select other than 'Max'

When the Class 800 fleet was new, a lot of drivers at my TOC correctly stated that the speeds out of London that were not attainable by the former fleet would easily be reached or exceeded with the Class 80x if not careful. I couldn't have agreed more, so my solution was to select approx 30-40% power and just let it run until posted line speeds exceed 95mph. Doesn't bust any speed limits or lose time.

When it comes to departing stations, the power stays in the lowest setting until I can be sure the train is clear of the platform. That first notch of power is still quicker than what the previous fleet could achieve from a stand, so there's no detriment to the timetable and it's smoother for all involved. The over-speeding incidents that have recently occurred on our network seem to have come off the back of drivers going straight to full power in situations that really don't warrant it.

Minimum power from Grantham for that set of points in an 80X would see you at our around 45mph. Not really a solution.

 

[Strat-tastic]

Out of interest I simulated this on Train Sim World, using max power just about straight away. I hit the junction at 56 mph.

 

[Railperf]

Seems to me there needs to be another repeater after the station platform and possibly a pair of TPWS grids to ensure compliance. Or else change the switch to one capable of 50-60mph so that a train leaving the station and taking the slow line can switch lines without having to slow down and remain at a slow speed.

The fact that the points are almost half a mile from the station where a driver concentrating on station duties might have forgotten about the diverging aspect - is an additional safety risk that thankfully didn't result in any fatalities.

Any reason why the signal when showing a diverging route isn't a flashing yellow or flashing double yellow?

 

[800001]

Seems to me there needs to be another repeater after the station platform and possibly a pair of TPWS grids to ensure compliance. Or else change the switch to one capable of 50-60mph so that a train leaving the station and taking the slow line can switch lines without having to slow down and remain at a slow speed.

The fact that the points are almost half a mile from the station where a driver concentrating on station duties might have forgotten about the diverging aspect - is an additional safety risk that thankfully didn't result in any fatalities.

Any reason why the signal when showing a diverging route isn't a flashing yellow or flashing double yellow?

Is the signal not visible on the platform at Grantham? So even if stationary the signal is checked by driver before departing? If it is, then surely the route indicator is also visible?

 

[endofthetunnel]

The signal is at the end of the platform.
So a stopping service is basically unaffected by approach released aspects as it’s stopping there anyway and the signal has stepped up to a green (with route indicator) by the time it is ready for departure.

 

[Tomnick]

Any reason why the signal when showing a diverging route isn't a flashing yellow or flashing double yellow?

Because that's not how flashing yellow sequences work. The signal in rear of the junction signal would be the one displaying a flashing (single) yellow, but in any case the differential in permissible speed between the straight route and diverging route is too great in this case to permit flashing yellows to be used. Even if they were, the same situation could arise if the route wasn't set until the train was in the platform.

 

[Meerkat]

As an outsider the suggestion above for a postliminary route indicator seems the simplest - are there any downsides apart from cost?

 

[AJD]

Minimum power from Grantham for that set of points in an 80X would see you at our around 45mph. Not really a solution.

I did this yesterday and hit it around 32mph in a lightly loaded 801. Granted it's still faster than the points speed, but going straight into full power from a stand wasn't/isn't really necessary and ended up making this wayyy worse.

There's definitely a trend where more and more drivers don't respond well to anything other than greens and main/fast lines. The idea of not learning slow lines, lesser used junctions and alternative routes because few trains are booked over them has become increasingly normalised. Personally, this is the stuff I feel we get paid for and while it's true not many trains use the slows, they're really not in the slightest bit difficult to learn. Can't be dealing with laziness when we get paid what we do.

I'm a 15 year driver, so not exactly old school by any means but when I was trained the onus was on the driver to really understand their signals and route. Blame culture has diluted that and, more than ever, it's now the signallers fault for someone having a SPAD as they didn't set the route and you seldom stop at that particular signal. Or it's the layout that's to blame for the recent Peterborough speeding incidents, wheras the layout is actually a very good one if a little time is spent studying it.

Just my 2p worth and no doubt some will disagree, but current trends do make for worrying reading.

 

[Unobrow]

I did this yesterday and hit it around 32mph in a lightly loaded 801. Granted it's still faster than the points speed, but going straight into full power from a stand wasn't/isn't really necessary and ended up making this wayyy worse.

There's definitely a trend where more and more drivers don't respond well to anything other than greens and main/fast lines. The idea of not learning slow lines, lesser used junctions and alternative routes because few trains are booked over them has become increasingly normalised. Personally, this is the stuff I feel we get paid for and while it's true not many trains use the slows, they're really not in the slightest bit difficult to learn. Can't be dealing with laziness when we get paid what we do.

I'm a 15 year driver, so not exactly old school by any means but when I was trained the onus was on the driver to really understand their signals and route. Blame culture has diluted that and, more than ever, it's now the signallers fault for someone having a SPAD as they didn't set the route and you seldom stop at that particular signal. Or it's the layout that's to blame for the recent Peterborough speeding incidents, wheras the layout is actually a very good one if a little time is spent studying it.

Just my 2p worth and no doubt some will disagree, but current trends do make for worrying reading.

Now this, I totally agree with. As a freight driver, we rarely use the fast lines but obviously need to know the same things that have been mentioned here. Obviously there’s less risk to us going this way rather than the fast to slow routes, but it’s the same bottom line. RTC for example, whilst not my thing in the format it’s promoted (I still do it, but the way that suits me), has its big pluses. Constantly talking to/reminding yourself what was last and is coming up can obviated that degree of concentration lapse.

At the end of the day, it’s the drivers responsibility to learn and know the routes they sign. One of our union reps reminds us regularly of this, and for good reason. Especially the more you sign and therefore could drive less regularly.

Now before I’m pounced upon, I’m certainly not holier than thou. I’ve suffered the usual under load and switching off in this job and my previous driving career, but I’ll always admit to that and be honest with myself.

 

[Stossgebet]

How about installing preliminary route indicators with the speed illuminated within the PRI box, appropriate for the speed of the diverging route that is set. It would not take away the possibilty of an overspeed at the points. But it will reduce the risk, by giving the driver an illuminated speed to abide by.

 

[Nicholas Lewis]

I did this yesterday and hit it around 32mph in a lightly loaded 801. Granted it's still faster than the points speed, but going straight into full power from a stand wasn't/isn't really necessary and ended up making this wayyy worse.

There's definitely a trend where more and more drivers don't respond well to anything other than greens and main/fast lines. The idea of not learning slow lines, lesser used junctions and alternative routes because few trains are booked over them has become increasingly normalised. Personally, this is the stuff I feel we get paid for and while it's true not many trains use the slows, they're really not in the slightest bit difficult to learn. Can't be dealing with laziness when we get paid what we do.

I'm a 15 year driver, so not exactly old school by any means but when I was trained the onus was on the driver to really understand their signals and route. Blame culture has diluted that and, more than ever, it's now the signallers fault for someone having a SPAD as they didn't set the route and you seldom stop at that particular signal. Or it's the layout that's to blame for the recent Peterborough speeding incidents, wheras the layout is actually a very good one if a little time is spent studying it.

Just my 2p worth and no doubt some will disagree, but current trends do make for worrying reading.

Well said and personally I felt the Peterborough RAIB reports didn't get to the heart of the underlying cause and you have given them plenty of pointers here for this investigation.

 

[Annetts key]

In most other industries on anything safety critical you have a 'plant control system' and a 'safety system' The plant control system runs the plant effectivley most of the time, and may make decisions based on human or other machine input. The safety system sits there monitoring critical plant parameters, separately and simply, and takes no information from the plant control system. If any of the critical operating parameters exceed certain thresholds then the safety system shuts the process down safely and independently. In rail terms the 'plant control system' is the driver, but the safety system does not have appeared to developed to the same exent on railways.

The technology was developed many years ago after the collision and crash at Clapham Junction. One of the two pilot systems was fitted to parts of the GWML. It's called ATP (automatic train protection) but is now considered obsolete (although the GWML system is maintained and is operational), so would not be considered for new installations.

ETCS/ERTMS is the current technology. But the roll out of this has been delayed due to various reasons.

ATP was not installed widely across the network because it was considered to be too expensive by the then government. And some in the railway didn't want it. The view of some was that it limited capacity.

Instead, after the joint investigation into safety systems after the Ladbroke Grove and Southall crashes, TPWS was installed. But this is a more limited system and although effective at mitigating or preventing some crashes, it doesn't provide the more comprehensive protection of ATP or ETCS/ERTMS.

 

[anothertyke]

I did this yesterday and hit it around 32mph in a lightly loaded 801. Granted it's still faster than the points speed, but going straight into full power from a stand wasn't/isn't really necessary and ended up making this wayyy worse.

There's definitely a trend where more and more drivers don't respond well to anything other than greens and main/fast lines. The idea of not learning slow lines, lesser used junctions and alternative routes because few trains are booked over them has become increasingly normalised. Personally, this is the stuff I feel we get paid for and while it's true not many trains use the slows, they're really not in the slightest bit difficult to learn. Can't be dealing with laziness when we get paid what we do.

I'm a 15 year driver, so not exactly old school by any means but when I was trained the onus was on the driver to really understand their signals and route. Blame culture has diluted that and, more than ever, it's now the signallers fault for someone having a SPAD as they didn't set the route and you seldom stop at that particular signal. Or it's the layout that's to blame for the recent Peterborough speeding incidents, wheras the layout is actually a very good one if a little time is spent studying it.

Just my 2p worth and no doubt some will disagree, but current trends do make for worrying reading.

I can see all that in relation to the two incidents at Peterboro because AIUI they were confronted with a relatively unusual manoevre and failed to expect the unexpected. In the Grantham case, you have a train which is scheduled to do that every morning if the train behind is on time. So I'm not sure it is about route learning or laziness, more likely to be 'had a blank' and then the question is why. Just my 2p, not a driver.

 

[43066]

I can see all that in relation to the two incidents at Peterboro because AIUI they were confronted with a relatively unusual manoevre and failed to expect the unexpected. In the Grantham case, you have a train which is scheduled to do that every morning if the train behind is on time. So I'm not sure it is about route learning or laziness, more likely to be 'had a blank' and then the question is why. Just my 2p, not a driver.

My impression from the reports was that, although various factors were involved, lack of route knowledge was a major part of the Lumo incident. Of course you’d also generally expect a driver who was unsure to slow down if they received a feather they were unfamiliar with.

In the Grand Central incident the driver apparently hadn’t realised he’d received the feather and thought he was taking the mainline, so that points more to a lack of NTS/situational awareness.

 

[Clarence Yard]

The chronology is important here.

The RAIB put a lot of emphasis on training in the first incident, something that FG were not entirely happy with because they thought the signal layout/spacing was a major factor. It was leading the driver to a false conclusion.

Then came the second incident and it became obvious to all that was a wider issue here. Now we have another (nearby) ECML location where situational awareness with an 80x has been an issue.

 

[Haywain]

Now we have another (nearby) ECML location where situational awareness with an 80x has been an issue.

But this was from a standing start at a station call where the driver should have had the signal in his eye line for a couple of minutes. This was much more likely to have been down to the driver being distracted.

 

[AJD]

The chronology is important here.

The RAIB put a lot of emphasis on training in the first incident, something that FG were not entirely happy with because they thought the signal layout/spacing was a major factor. It was leading the driver to a false conclusion.

Then came the second incident and it became obvious to all that was a wider issue here. Now we have another (nearby) ECML location where situational awareness with an 80x has been an issue.

I remember this being discussed at work. Of course FG weren't entirely happy with the RAIB mentioning their training because it negatively puts them in the spotlight. They argued that their training requirements were met, but if you're only setting 175 handling hours, and think that's adequate, then I'm not surprised the RAIB were critical. As I've mentioned a few times the layout is sound and all it takes is a little study to understand why things are the way they are.

Personally, I feel the mention of the layout gave FG and others involved the excuse to apportion blame elsewhere.

 

[HSTEd]

This has essentially become an argument about procedural vs engineered controls on risk.
Huge numbers of signals are interpreted by drivers every day. No matter how good the training is, eventually mistakes will be made.

We've now apparently encountered a risk that is not well mitigated by existing train protection arrangements

I think there is an argument that the current arrangements are inadequate for trains with the performance that 8xx class units are capable of.

 

[800001]

I remember this being discussed at work. Of course FG weren't entirely happy with the RAIB mentioning their training because it negatively puts them in the spotlight. They argued that their training requirements were met, but if you're only setting 175 handling hours, and think that's adequate, then I'm not surprised the RAIB were critical. As I've mentioned a few times the layout is sound and all it takes is a little study to understand why things are the way they are.

Personally, I feel the mention of the layout gave FG and others involved the excuse to apportion blame elsewhere.

Surely the amount of handling hours needs to be an agreed amount agreed by RAIB?