RAIB report: "Near miss between a passenger train and cars at Norwich Road level crossing"

[Jim Martin]

Strewth: https://www.gov.uk/government/news/report-152020-near-miss-between-a-passenger-train-and-cars-at-norwich-road-level-crossing

 

I've searched, but this doesn't seem to be on the site yet. You'd like to think that this would highlight why "leaves on the line" are no joke; but I doubt that's how it'll come across.

 

Jim

 

[DY444]

The effect of leaves on the line I can understand and forgive. 

 

Less so the apparent lack of appreciation of the need to set the loss of signal timeout to its maximum value, the haphazard last minute approach to this at other installations, and the inconsistent application of the back up system using treadles.  Looks like a lack of understanding of the system, its risks and their mitigation, a lack of a consistent process for its design and installation across the network and a failure to review all of those in light of the previous incident with the Class 67.  If things had played out only fractionally differently then 2 or 3 road vehicles could have been struck.   Doesn't make pretty reading for an organisation obsessed with voluminous safety processes.

 

Also Stadler's lack of understanding of how the sanding system on its own train design was set up doesn't reflect very well on them either. 

Edited December 14, 2020 by DY444

[duncan]

I know the car drivers were in the right to cross, but did they not glance to the side when about to cross - don't most people ?  Were the train lights not clearly visible, as it was after dark ?  Or am I just IAM trained ?

[DY444]

1 minute ago, duncan said:

I know the car drivers were in the right to cross, but did they not glance to the side when about to cross - don't most people ?  Were the train lights not clearly visible, as it was after dark ?  Or am I just IAM trained ?

 

At the crossing not far from me you can't see up the line at all until you're actually on the crossing itself. 

 

In this case they were stationary at the barriers which then rose so they moved forward onto the crossing as most people would.  I don't know for sure but it's certainly possible that had the driver who had the very near miss proceeded more cautiously with a view to checking the line he might well have been hit. 

[Michael Hodgson]

not sure of current practice, but AHB crossings used to have the track circuits duplicated by strike-in and strike-out treadles to prevent this sort of thing. 

2 hours ago, duncan said:

I know the car drivers were in the right to cross, but did they not glance to the side when about to cross - don't most people ?  Were the train lights not clearly visible, as it was after dark ?  Or am I just IAM trained ?

 

Most drivers don't. The highway code never required it (though I haven't looked at it since it cost 6d).  

You often cant see very far along the line from a car anyway, and road traffic may often be 50 mph or more. 

If you slow down too much you'll cause a collision with other road users.

[phil-b259]

Dare I suggest this is another example of the Railtrack ethos of wanting to use novel signalling systems*  because they are cheep** rather than more traditional (and Robust ones) that are more expensive and maintenance heavy.

 

 

* Remember the Portsmouth (Siemens but before they bought Westinghouse) Manchester South (Ansaldo), Horsham (Bombardier) resignalling fiascos. Railtrack deliberately excluded the established companies of Westinghouse and Alston citing them being an expensive duopoly and Railtrack boses thus pumped for a host of cheep newcomers with untested technology that was frankly disastrous when used in the UK. Portsmouth encountered serious delays, Manchester south had to be de-scoped / massively curtailed and Horsham abandoned when it became clear these new entrants simply didn’t have the know-how or technology to actually deliver a decent signalling system. Yes Westinghouse and Alston may be expensive - but their systems are usually very good having benefited from decades of experience in delivering UK signalling.

 

 

* *Traditional treadles are a hindrance to track machines - many have been damaged by tampers or grinders in the past while the need to duplicate them to cater for multiple approach speeds introduces further wiring complexities + associated relays, all of which introduce yet more potential points of failure.

 

 

 

 

Edited December 15, 2020 by phil-b259

[phil-b259]

6 hours ago, Jim Martin said:

Strewth: https://www.gov.uk/government/news/report-152020-near-miss-between-a-passenger-train-and-cars-at-norwich-road-level-crossing

 

I've searched, but this doesn't seem to be on the site yet. You'd like to think that this would highlight why "leaves on the line" are no joke; but I doubt that's how it'll come across.

 

Jim

 

The original incident did have a thread going on the site from the time it happened. Unfortunately it started to deviate into speculation about the cause and as with other incidents where the RAIB are involved, the site moderators locked the thread to stop the speculation continuing.

[Jim Martin]

3 hours ago, duncan said:

I know the car drivers were in the right to cross, but did they not glance to the side when about to cross - don't most people ?  Were the train lights not clearly visible, as it was after dark ?  Or am I just IAM trained ?

 

I had a look at the crossing on Streetview. Because the track and the road are at quite an angle, the stop lines are well back from the railway tracks. There's very little view along the tracks until you're on them.

 

This is the crossing (Google says that the road here is called Salhouse Road; it becomes Norwich Road almost immediately out-of-shot to the left). The train would have been coming from the bottom of the picture towards the top:

 

This is the view of the tracks for a car stopped at the barrier, for the vehicle crossing from left to right in front of the train:

 

And this is for a vehicle stopped at the barrier crossing from right to left (i.e. train coming from the left), which would seem to have the better view towards the approaching train:

 

Jim

Edited December 14, 2020 by Jim Martin
To clarify one of the picture captions

[phil-b259]

3 hours ago, DY444 said:

The effect of leaves on the line I can understand and forgive. 

 

Less so the apparent lack of appreciation of the need to set the loss of signal timeout to its maximum value, the haphazard last minute approach to this at other installations, and the inconsistent application of the back up system using treadles.  Looks like a lack of understanding of the system, its risks and their mitigation, a lack of a consistent process for its design and installation across the network and a failure to review all of those in light of the previous incident with the Class 67.  If things had played out only fractionally differently then 2 or 3 road vehicles could have been struck.   Doesn't make pretty reading for an organisation obsessed with voluminous safety processes.

 

Also Stadler's lack of understanding of how the sanding system on its own train design was set up doesn't reflect very well on them either. 

 More like Railtrack didn’t know what they were doing!

 

I note that the later use of these predictors on the Bletchley to Bedford line (which was a Network Rail project IIRC) did see more attention paid as the ‘keep the barriers down if the train disappears) timer was set to the maximum allowed (99 seconds) rather than kept at the default 16 seconds as per the Cromer line ones.

[Jim Martin]

I take it that the idea of the predictors is to tailor the amount of time the barriers are down to the speed of the train; so you don't get the situation where the barriers close then nothing happens for ages because the approaching train's going well below line speed?

 

Jim

[DY444]

9 hours ago, phil-b259 said:

 More like Railtrack didn’t know what they were doing!

 

I note that the later use of these predictors on the Bletchley to Bedford line (which was a Network Rail project IIRC) did see more attention paid as the ‘keep the barriers down if the train disappears) timer was set to the maximum allowed (99 seconds) rather than kept at the default 16 seconds as per the Cromer line ones.

 

I think Railtrack's incompetence and, in some cases, negligence when it comes to engineering is well understood and which, very sadly, is the unwritten postscript on a number of headstones.

 

What bothers me from this case is the apparent inability of NR to put 2 and 2 together regarding crossings that use predictors relying on detecting the presence of rail borne electrical signals.  It has been known for decades that BR second generation DMUs and their successors can't be relied on to operate track circuits properly in all conditions hence the fitting of TCAs to such units.  Yet here we have a system which uses far lower level signals than track circuits which had no backup at all apart from a timer which as this case shows is safety critical but not everyone seemed to realise that and no one thought to check across the network after incidents elsewhere.  

 

Wrt treadles being a nuisance to track maintenance, is it beyond the wit of man to devise a treadle based on an axle counter head which might be less of a nuisance?

Edited December 15, 2020 by DY444

[spamcan61]

8 hours ago, Jim Martin said:

I take it that the idea of the predictors is to tailor the amount of time the barriers are down to the speed of the train; so you don't get the situation where the barriers close then nothing happens for ages because the approaching train's going well below line speed?

 

Jim

Yes, see paragraph 38 of the report onwards. 

 

Personally, as someone with only a passing interest in 1:1 scale trains and no insider knowledge of the Rail industry, I find these RAIB reports very well written. Everything is clearly stated, well laid out; of course they are technical in places, but no more technical than they need to be. A model of clear concise reports in my opinion.

 

From my day job involving complex electronic systems I do start worrying when I see settings left at default.

[TheQ]

They were interviewing  (recorded) on the radio, a NR representative as I passed by the end of that road this morning..  Lots of what they've done to stop it happening again.. cut back the hedges and trees, added a treadle counter, to monitor the trains... But right at the end of the interview you got the real reason it happened.. 

 

They stopped the weekend track cleaning trains, they wouldn't say why, but you can guess it was a bean counter somewhere trying to save money..

 

The weekend track cleaning trains have restarted..

[phil-b259]

4 minutes ago, TheQ said:

They were interviewing  (recorded) on the radio, a NR representative as I passed by the end of that road this morning..  Lots of what they've done to stop it happening again.. cut back the hedges and trees, added a treadle counter, to monitor the trains... But right at the end of the interview you got the real reason it happened.. 

 

They stopped the weekend track cleaning trains, they wouldn't say why, but you can guess it was a bean counter somewhere trying to save money..

 

The weekend track cleaning trains have restarted..

 

Relying on track cleaning trains is not a reasonable long term solution though - they can get cancelled for a whole host of legitimate reasons including equipment failure, staff shortages or engineering works. As the RAIB also point out the coverage will vary year on year as like the flu vaccine, their deployment (and frequency) is tweaked year on year to reflect where railhead conditions are going to be worst.

 

What seems to have happened here is that Railhead Treatment trains actually covered up the fact that the level crossing equipment was inherently unsafe from the beginning. A 'perfect storm' of new rolling stock, dry weather and the lack of railhead treatment combined to allow that fundamental flaw to actually manifest itself.

 

While its obviously disturbing we came so close to a serious accident, had the railhead treatment trains not ceased to run at weekends then the ticking time bomb of a unsafe level crossing installation would have remained hidden!

 

 

[TravisM]

9 hours ago, Jim Martin said:

 

I had a look at the crossing on Streetview. Because the track and the road are at quite an angle, the stop lines are well back from the railway tracks. There's very little view along the tracks until you're on them.

 

This is the crossing (Google says that the road here is called Salhouse Road; it becomes Norwich Road almost immediately out-of-shot to the left). The train would have been coming from the bottom of the picture towards the top:

 

This is the view of the tracks for a car stopped at the barrier, for the vehicle crossing from left to right in front of the train:

 

And this is for a vehicle stopped at the barrier crossing from right to left (i.e. train coming from the left), which would seem to have the better view towards the approaching train:

 

Jim

The high trees and other vegetation wouldn’t have helped as it restricted the view of not only the car drivers but the train driver as well.  I’ve always advocated that trees and suchlike should be cut back so both the road users and railway have a clear view of the crossing.

 

Slightly off topic but wasn’t vegetation the cause of a fatality on the far north line a few years ago as it had obscured the flashing lights and as the crossing had no barriers, the elderly driver just drove onto the crossing oblivious to the train?

[DY444]

22 minutes ago, TheQ said:

They were interviewing  (recorded) on the radio, a NR representative as I passed by the end of that road this morning..  Lots of what they've done to stop it happening again.. cut back the hedges and trees, added a treadle counter, to monitor the trains... But right at the end of the interview you got the real reason it happened.. 

 

They stopped the weekend track cleaning trains, they wouldn't say why, but you can guess it was a bean counter somewhere trying to save money..

 

The weekend track cleaning trains have restarted..

 

That's not the real reason it happened.  The real reason it happened was because the crossing design/configuration failed to adhere to one of the most basic cornerstones of railway operation namely that if something fails it should fail safe.

Edited December 15, 2020 by DY444

[phil-b259]

9 hours ago, Jim Martin said:

I take it that the idea of the predictors is to tailor the amount of time the barriers are down to the speed of the train; so you don't get the situation where the barriers close then nothing happens for ages because the approaching train's going well below line speed?

 

Jim

 

Thats the idea - if the line speeds between different types of traction (or the speeds between stopping and non stopping trains are significantly different) then you need to effectively duplicate the amount of treadles and quite possibility install extra signals to compensate. This increases costs significantly and introduces more things to fail.

 

Given the disregard the UK population seem to show to half barrier crossings* and their propensity to weave round the barriers rather than wait the concept of using predictors to tailor the crossing operation to the approaching train is however a noble one.

 

 

* I understand that NR and the ORR have put in place a ban on any more AHBs being created in the UK precisely because the public cannot be trusted to use them properly - and UK law means its the railway that gets the blame unlike in continental Europe!

[phil-b259]

8 minutes ago, jools1959 said:

The high trees and other vegetation wouldn’t have helped as it restricted the view of not only the car drivers but the train driver as well.  I’ve always advocated that trees and suchlike should be cut back so both the road users and railway have a clear view of the crossing.

 

Slightly off topic but wasn’t vegetation the cause of a fatality on the far north line a few years ago as it had obscured the flashing lights and as the crossing had no barriers, the elderly driver just drove onto the crossing oblivious to the train?

 

On the other hand there have been cases where clear visibility means drivers are tempted to try and 'beat' the train.

 

I'm told that this was an issue with Hixon crossing in Staffordshire - the way the railway was clearly visible for a significant distance as you approached from the A51 meant there were a number of incidents of this right up until a bridge was put in during the late 1990s WCML modernisation.

 

Where level crossing equipment is installed users (and train drivers) should be able to rely on that - the fact that Railtrack allowed a substandard and inherently unsafe crossing system to be installed under their watch should be a wake up call to those in power who are tempted to cut costs by the use of novel equipment.

[Michael Hodgson]

Let's keep this in proportion.  Now that I've read the RAIB report it's worth noting that 

• there's 25,000 crossings globally using these predictors, and they've been around for 20 years, so they're not new unproven technology 
• there wasn't actually a collision and nobody got hurt; near miss detection and investigation is worthwhile as it is part of the industry's accident avoidance process - you don't get that for road traffic
• treadles have now been fitted retrospectively to all crossings with these predictors, so that particular problem can't happen again
• the prediction concept is itself a safety enhancement in that it is likely to reduce the incidence of motorists deciding that the crossing has failed and unlawfully dodging round the barriers.  It also reduces delay and inconvenience to road traffic
• the RAIB report also identifies and addresses other management failings as regards safety

An interesting issue to come out of the report is that nearly all of the movements on that line are new rolling stock with unworn wheels.   When that service was provided by clapped out trains with worn wheels,  it was less likely to happen!  The RAIB report stopped short of recommending worn axles be fitted to new trains though 

 

[iands]

15 hours ago, phil-b259 said:

Dare I suggest this is another example of the Railtrack ethos of wanting to use novel signalling systems*  because they are cheep** rather than more traditional (and Robust ones) that are more expensive and maintenance heavy.

 

 

* Remember the Portsmouth (Siemens but before they bought Westinghouse) Manchester South (Ansaldo), Horsham (Bombardier) resignalling fiascos. Railtrack deliberately excluded the established companies of Westinghouse and Alston citing them being an expensive duopoly and Railtrack boses thus pumped for a host of cheep newcomers with untested technology that was frankly disastrous when used in the UK. Portsmouth encountered serious delays, Manchester south had to be de-scoped / massively curtailed and Horsham abandoned when it became clear these new entrants simply didn’t have the know-how or technology to actually deliver a decent signalling system. Yes Westinghouse and Alston may be expensive - but their systems are usually very good having benefited from decades of experience in delivering UK signalling.

 

 

* *Traditional treadles are a hindrance to track machines - many have been damaged by tampers or grinders in the past while the need to duplicate them to cater for multiple approach speeds introduces further wiring complexities + associated relays, all of which introduce yet more potential points of failure.

 

 

 

 

I agree Phil. I was with Railtrack from day one (1st April 1994). Very early on in its short life, at one of the regular meetings with the bosses, it was the view of "Railtrack Corporate" that the "Standards" were not there just to be followed/adhered to, but to be "challenged". The message being that we, as engineers, should rigorously examine ways of improving the cost/benefit ratio at every opportunity (in other words, reducing costs). Whilst I was on the telecoms side rather than the signalling side, quite a few of us still felt somewhat uncomfortable with this approach, especially as there were quite a few "new starters" from outside the industry appointed to higher positions in management/procurement/contracts that loved this sort of approach and became an almost daily mantra. Sadly, and tragically, serious lessons were learnt before things improved again. However, as clearly seen from this incident, there is still a "Railtrack legacy" to be dealt with. 

[Reorte]

38 minutes ago, iands said:

I agree Phil. I was with Railtrack from day one (1st April 1994). Very early on in its short life, at one of the regular meetings with the bosses, it was the view of "Railtrack Corporate" that the "Standards" were not there just to be followed/adhered to, but to be "challenged". The message being that we, as engineers, should rigorously examine ways of improving the cost/benefit ratio at every opportunity (in other words, reducing costs). Whilst I was on the telecoms side rather than the signalling side, quite a few of us still felt somewhat uncomfortable with this approach, especially as there were quite a few "new starters" from outside the industry appointed to higher positions in management/procurement/contracts that loved this sort of approach and became an almost daily mantra. Sadly, and tragically, serious lessons were learnt before things improved again. However, as clearly seen from this incident, there is still a "Railtrack legacy" to be dealt with. 

 

OK I'm speaking as an outside here but that doesn't sound like a fundamentally flawed approach - it only is if it's been implemented with the bias firmly set towards reducing costs and nothing else. A properly open environment would mean that the changed standards could be challenged too - nothing should be sacred (which doesn't mean changed lightly either), and I doubt even the most old-fashioned change-hating curmudgeon would ever claim everything's perfect and can't be improved. For that environment to work though you certainly need the experience in the right place, which from what you say sounds like a lacking component.

[phil-b259]

6 hours ago, Michael Hodgson said:

Let's keep this in proportion.  Now that I've read the RAIB report it's worth noting that 

• there's 25,000 crossings globally using these predictors, and they've been around for 20 years, so they're not new unproven technology.

 

They were unproven in the UK which is the point!

 

I refer you to the signalling systems from Ansaldo, etc - they were ‘well proven’ in their home countries but were incapable of working properly when applied to the British signalling environment without extensive re-work and de-scoping.

 

The rules and many of the fundamental concepts the UK signalling uses are NOT the same as other countries - some like Germany are based round ‘speed signalling’ (where the signals basically tell drivers what speed to go at) and not ‘route based signalling’ (where signals tell the drivers which route they are taking and not the speed they should be travelling at).

 

Consequently a German interlocking system will NOT work properly transplanted to the UK regardless of how many have been installed in Germany, Austria, etc. It MAY work OK after modifications have been made - but said modifications do not come cheap and do need input from experienced UK signalling experts - not just accountants to verify a forgiven system is cheaper than a Westinghouse / Alston product.
 

Yes it’s pleasing that the flaw in the use of predictors in the UK has been fixed - but the point is it shouldn’t have been overlooked in the first place! As the RAIB report makes clear we came within half a second of a major incident, most likely involving loss of life, infrastructure damage and a loss of public confidence in the railways - the last of which will ironically push more people back onto the far more dangerous roads.

[St. Simon]

1 hour ago, phil-b259 said:

The rules and many of the fundamental concepts the UK signalling uses are NOT the same as other countries - some like Germany are based round ‘speed signalling’ (where the signals basically tell drivers what speed to go at) and not ‘route based signalling’ (where signals tell the drivers which route they are taking and not the speed they should be travelling at).

 

Hi Phil,

 

At risk of going off topic, but the use of speed signalling or route based signalling isn't the same as the interlocking principles.

 

If we were to convert from a route based lineside signalling to speed based lineside signalling, the only parts of the interlocking which are affected is a small part of the aspect level of the interlocking, effectively the aspect sequence data. The route calling, locking and release data would remain fundamentally the same. It's only when you getting into system that requires two way communication between interlocking and train were the interlocking principles fundamentally change.

 

The differences between European and UK CBI interlockings (today at least) are more to do with what is written in the interlocking data and also the number of processors which are required (the UK uses 2 out of 3, where as some other European countries require 2 out of 2), but in theory, European and UK CBI interlockings can be swapped with not too much effort (relatively). 

 

 

Simon

 

[phil-b259]

39 minutes ago, St. Simon said:

 

Hi Phil,

 

At risk of going off topic, but the use of speed signalling or route based signalling isn't the same as the interlocking principles.

 

If we were to convert from a route based lineside signalling to speed based lineside signalling, the only parts of the interlocking which are affected is a small part of the aspect level of the interlocking, effectively the aspect sequence data. The route calling, locking and release data would remain fundamentally the same. It's only when you getting into system that requires two way communication between interlocking and train were the interlocking principles fundamentally change.

 

The differences between European and UK CBI interlockings (today at least) are more to do with what is written in the interlocking data and also the number of processors which are required (the UK uses 2 out of 3, where as some other European countries require 2 out of 2), but in theory, European and UK CBI interlockings can be swapped with not too much effort (relatively). 

 

 

Simon

 

However the point that differences exist is still valid and the procurement of non UK developed signalling solutions will require alterations to suit the UK enviroment regardless of how well overseas designs may perform on their home turf.
 

As it happens, the changes needed to make the level crossing predictors align with UK signalling practice may have been minor, but changes were undoubtably needed (and ignored by Railtrack) or this thread wouldn’t exist.

 

However in the cases of the three Railtrack resignaliing schemes I mentioned upthread, the alterations were so masive that they significantly impacted the delivery of the schemes such that one (EbiLock) had to be completely junked and replaced with the well proven BR SSI equipment throughout.

 

Network Rail does seem to be more aware of this these days and has basically gone back to using Siemens (I.e. Westinghouse) and Alstom as the principle suppliers of signalling technology which should avoid the mistakes of the Railtrack era - but at a higher costs due to the effective ‘duopoly’.

 

 

Edited December 15, 2020 by phil-b259

[Edwin_m]

22 hours ago, Michael Hodgson said:

You often cant see very far along the line from a car anyway, and road traffic may often be 50 mph or more. 

If you slow down too much you'll cause a collision with other road users.

There's also the risk if slowing or stopping that a driver will get into the wrong gear and stall on the crossing itself.  

21 hours ago, Jim Martin said:

I take it that the idea of the predictors is to tailor the amount of time the barriers are down to the speed of the train; so you don't get the situation where the barriers close then nothing happens for ages because the approaching train's going well below line speed?

That's the intention but the system alters the start of the crossing sequence, not the time the barriers are down, so as to keep the barrier-down time moreorless constant.