Welshpool crossing crash (22/06/20)

[DerekC]

Aircraft have been able to make approaches to runways in zero visibility, navigating by sole reference to GPS signals, for many decades now. The suggestion that GPS isn't reliable enough to be used for safety critical applications is absolute bunkum. Yes, the kind of GPS you get on a smartphone isn't good enough. But if you pay the right people enough money you can get kit that can be guaranteed to be sufficiently accurate and failsafe.

The simple truth of the matter is that not enough people have died, or will die, over the railway's stubbornness in this regard, for it to have become a priority.

Yes, there are places where GPS signals from satellites won't work. But it is no different to the implementation of GSM-R - where there is a will, there is a way.

The idea that aircraft fly around using GPS only is simply not true. Here's a link to an easy-to follow and up to date explanation of the systems used by commercial aircraft:

https://www.flightdeckfriend.com/2020/04/20/how-do-commercial-aircraft-navigate/

An extract from the article is as follows:

Commercial aircraft utilise a number of navigation systems to help guide the flight from point A to B. The systems are of varying accuracy, with GPS being the most accurate method (a bit like the Sat Nav in your car) but there are a number of other systems such as an internal navigation system which doesn’t use any external reference as well as some radio beacon receivers which are used for particular phases of flight such as the departure or arrival / approach segments.

GPS – Global Positioning System

This is one of the primary navigation sources and the aircraft is continuously attempting to monitor its GPS position. This is also typically the most accurate navigation system on most modern commercial aircraft, in some circumstances allowing the aircraft to perform manoeuvres down to an accuracy of 0.1 Nautical Miles.

However, due to its reliability, it can only be used for primary navigation during certain segments of the flight. To use GPS as the primary navigation reference for the Departure (Standard Instrument Departure or SID), or Arrival (Standard Terminal Arrival or STAR) or Approach (RNAV or RNP/GPS/GNSS Approach) specific approval for both the aircraft and airline is required.

GPS is susceptible to a number of limitations. First of all, there are GPS black spots where there may not be enough satellites in range to obtain a position. Clearly this is not acceptable if operating at the aircraft with sole reference to GPS when close to the ground. GPS is also subject to jamming or interruption due to a number of reasons, such as military conflict.

Should the aircraft GPS receivers fail, or if the GPS system/satellites were to go offline, the aircraft has suitable redundancy in-place to ensure it can still navigate to an acceptable level of accuracy.

IRS – Inertial Reference System (IRS)

The IRS (older versions are referred to as Inertial Navigation Systems) is a self contained system that is able to track the aircraft position with no external reference. It uses a combination of accelerometers and gyroscopes to calculate any movement and acceleration of the aircraft across any of its three axis.


Radio Aids

Radio beacons, normally located on land, send out radio beams which tell us the aircraft’s range and direction from that radio aid. This allows the aircraft’s computer systems to calculate the aircraft’s location. The more radio signals that can be detect, the more accurate the estimated position is.

These radio aids are typically referred to as Very High Frequency Omni-Directional Radio Range (VOR) with associated Distance Measuring Equipment (DME). Others include a Non-Directional Beacon (NDB) bit this purely provides a pointer towards the beacon rather than any direct radial or distance information.

Instrument Landing Systems (ILS) are used to help guide the aircraft into land.

 

[ForTheLoveOf]

The idea that aircraft fly around using GPS only is simply not true. Here's a link to an easy-to follow and up to date explanation of the systems used by commercial aircraft:

https://www.flightdeckfriend.com/2020/04/20/how-do-commercial-aircraft-navigate/

An extract from the article is as follows:

I am quite aware of the different navigation and approach systems aircraft use. You will note I did not suggest that GPS is the only means of navigation used, because it isn't. However you helpfully add to my point because, as your article states, "GPS is the most accurate method".

 

[LOM]

Because of that incident all UWC requests are now protected by reds no matter how long the user needs.

It may be the case where you are but that is not a national requirement.

 

[gsnedders]

My experience in GPS suggests otherwise, but it is several decades old (I was probably the first person in Europe to try it on a train) so I accept that may have changed. However nobody uses it in a railway safety-critical application as far as I'm aware.

Note that you don't need it perfect for a situation like level crossings, provided you have some knowledge of the quality of the data (e.g., computed accuracy, last measurement time, etc), because you only need to do better than "in a block that takes 20 minutes for the train to transverse", which means typically "it's somewhere within this 1km range" is an improvement.

 

[Llanigraham]

Would the signaller automatically know which (user worked crossing) telephone a call is coming in from? (i.e. Caller display?)

Yes they would. If I remember rightly, at Machy it displays on the screens and on the concentrator.
At my (manual) Box it displayed on the concentrator, and before that each crossing had it's own phone on the Block Shelf.

 

[DerekC]

I am quite aware of the different navigation and approach systems aircraft use. You will note I did not suggest that GPS is the only means of navigation used, because it isn't. However you helpfully add to my point because, as your article states, "GPS is the most accurate method".

Indeed the article does say that ""GPS is the most accurate method" but it also says " however, due to its reliability, it can only be used for primary navigation during certain segments of the flight" and explains why other systems are used to augment and confirm the GPS position and to allow the aircraft to operate safely if GPS fails. Similar issues apply to the use of GPS on railways only more so because (as somebody said) railways have to cope with tunnels, cuttings and reflections from nearby buildings, none of which apply to an aircraft.

Having said that I agree that railways have been extremely slow to use the technology in some areas.

I have a feeling this needs a thread of its own.

 

[Llanigraham]

Aircraft have been able to make approaches to runways in zero visibility, navigating by sole reference to GPS signals, for many decades now. The suggestion that GPS isn't reliable enough to be used for safety critical applications is absolute bunkum. Yes, the kind of GPS you get on a smartphone isn't good enough. But if you pay the right people enough money you can get kit that can be guaranteed to be sufficiently accurate and failsafe.

The simple truth of the matter is that not enough people have died, or will die, over the railway's stubbornness in this regard, for it to have become a priority.

Yes, there are places where GPS signals from satellites won't work. But it is no different to the implementation of GSM-R - where there is a will, there is a way.

How much would it cost to fit every signal box with the required instruments and displays? Remember that there is still much of the country that is signalled mechanically by people pulling levers from the Victorian era.

At our Box the rules were quite clear, if we had accepting a train from either of the bordering Boxes then NO-ONE got permission to cross until we had received Train out of Section from the next Box.

 

[edwin_m]

Indeed the article does say that ""GPS is the most accurate method" but it also says " however, due to its reliability, it can only be used for primary navigation during certain segments of the flight" and explains why other systems are used to augment and confirm the GPS position and to allow the aircraft to operate safely if GPS fails. Similar issues apply to the use of GPS on railways only more so because (as somebody said) railways have to cope with tunnels, cuttings and reflections from nearby buildings, none of which apply to an aircraft.

Having said that I agree that railways have been extremely slow to use the technology in some areas.

I have a feeling this needs a thread of its own.

Indeed, the issue is one of integrity rather than accuracy. It would be fair enough if it failed and told people it had failed, so they could revert to other methods in the meantime. The worst situation is if it fails but still appears to be working. This can't be ruled out unless the GPS itself on the train, and all the communications systems that report the GPS position back to the signaler, are of sufficient integrity.

 

[InOban]

Of course GPS doesn't give a clear position all the time, but it can be integrated with other data sources, eg speed and time, to a location which is more than accurate enough.
And of course you don't have level crossings in cuttings or tunnels!
The next upgrade to the radio signalling on the Far North and West Highland lines will integrate GPS so that the Box knows exactly where the train is, and can therefore authorise us of a UWC.

 

[edwin_m]

Of course GPS doesn't give a clear position all the time, but it can be integrated with other data sources, eg speed and time, to a location which is more than accurate enough.
And of course you don't have level crossings in cuttings or tunnels!
The next upgrade to the radio signalling on the Far North and West Highland lines will integrate GPS so that the Box knows exactly where the train is, and can therefore authorise us of a UWC.

As I posted above, the issue is not so much accuracy as integrity. And there are many places where a train can be in a cutting or tunnel less than a minute before passing over a gated crossing.

Interested if you could quote a source for more details of integration of GPS into RETB, as it seems to me it would be breaking new ground in system safety.

 

[Belperpete]

Of course GPS doesn't give a clear position all the time, but it can be integrated with other data sources, eg speed and time, to a location which is more than accurate enough.
And of course you don't have level crossings in cuttings or tunnels!
The next upgrade to the radio signalling on the Far North and West Highland lines will integrate GPS so that the Box knows exactly where the train is, and can therefore authorise us of a UWC.

You don't need to be in a cutting or tunnel to get a significant error in GPS positioning. Somewhere where a large mountain or hill, or a tall office block, blocks sufficient satellite feeds and causes reflected feeds from others, can do it. At home, I am quite regularly told I am in a village three miles away.

As you say, the important thing is to integrate it with other location methods, so that the system knows when to ignore what the GPS is saying.

 

[Belperpete]

Note that you don't need it perfect for a situation like level crossings, provided you have some knowledge of the quality of the data (e.g., computed accuracy, last measurement time, etc), because you only need to do better than "in a block that takes 20 minutes for the train to transverse", which means typically "it's somewhere within this 1km range" is an improvement.

That is true, except for the 1km or so immediately around the crossing, where it is very important to know which side of the crossing the train is. A crossing user who spots a train approaching is likely to wait until the train has passed, and will then be straight onto the signaller. However, the signaller can't rely on the user doing so. Yes, the signaller can ask if the train has passed, but that makes it look like the signaller doesn't know where his trains are, and I seem to recall at least one accident where the road user confirmed that the train had passed, only he was talking about a previous train.

 

[Tomnick]

Because of that incident all UWC requests are now protected by reds no matter how long the user needs.

The risk with that, on a relatively busy line with relatively long sections and relatively well used UWCs (been there!), is that it gets much harder to find a suitable margin so the crossing user has to wait much longer, and in turn is more likely to just take matters into their own hands.

 

[Llanigraham]

The risk with that, on a relatively busy line with relatively long sections and relatively well used UWCs (been there!), is that it gets much harder to find a suitable margin so the crossing user has to wait much longer, and in turn is more likely to just take matters into their own hands.

Exactly the problem we had, and it led to abuse by certain users, but good explanations and building good relationships with regular users atleast reduced their "annoyance".
It was noticeable that certain signallers had less problems than others!

 

[Tom Quinne]

Indeed that is a risk, you just have to trust the user to wait until you Authorise them to across.

The bigger risk to my freedom which is the most important thing in my eyes, is I don’t put collars on and replace to red the signals, they cross fall over, I assume it clears which is the official SOP and they get taken out on the deck.

 

[Tom Quinne]

It may be the case where you are but that is not a national requirement.

Thats correct, assume crossing clear and don’t replace signals if under 3 minutes is the national SOP.
Our rule was a link only instruction, then box instruction after the north wales incident.

I always replaced and collared, and maintained until they called back with crossing clear.
Safety, safety, safety....until it starts delaying trains aka costing money.....not on my panel.

 

[DerekC]

That is true, except for the 1km or so immediately around the crossing, where it is very important to know which side of the crossing the train is. A crossing user who spots a train approaching is likely to wait until the train has passed, and will then be straight onto the signaller. However, the signaller can't rely on the user doing so. Yes, the signaller can ask if the train has passed, but that makes it look like the signaller doesn't know where his trains are, and I seem to recall at least one accident where the road user confirmed that the train had passed, only he was talking about a previous train.

There is kit available these days which integrates the GPS signal with inertial navigation and video sensors to provide a highly dependable track-precise position. Here's a link to something that does just that, specifically for railways (no, I don't have shares!)

http://www.rdsintl.com/train-positioning-system/1-navigation-engine/

t the heart of our Train Positioning System is the RDS Navigation Engine software, which can take inputs from multiple sensors to calculate the train’s track-precise position, and then output position data for use in rail applications such as passenger information systems, driver advisory systems (DAS), condition monitoring and fall-back signalling systems.

The Navigation Engine software was developed for our own Video Train Positioning System, but can interface with multiple sensors (wheel odometers, GNSS, inertial sensors), and information sources (Train Describer feeds, track maps), allowing a modular approach to train positioning system deployments and installations.

For applications that require a high-accuracy position the system is built around VTPS (and) supplemented by an inertial sensor, train describer feed and/or VTPS (with sign reader feature).

Applications of track-level position feeds include driver support systems (DSS), driver advisory systems (DAS), remote condition monitoring and selective door opening systems. A high integrity system configuration could also support low-cost and fall-back signalling

This is the kind of thing that would be very suitable for determining accurate train position for user worked crossings.

 

[Tomnick]

Indeed that is a risk, you just have to trust the user to wait until you Authorise them to across.

The bigger risk to my freedom which is the most important thing in my eyes, is I don’t put collars on and replace to red the signals, they cross fall over, I assume it clears which is the official SOP and they get taken out on the deck.

It is a difficult one, balancing that risk against the risk of them getting fed up and crossing anyway! You should be able to trust crossing users but I can understand their irritation if, for example, they’re regularly having to wait for three or four trains each several minutes apart (more so at those crossings that are used several times daily) and my concern as a driver would then be that they don’t bother phoning at all because they’re always made to wait an inexplicably (to them) long time.

Your actions should never be called into question in that situation though?

On the other hand, we had one regular crossing user who would always ring back whether you asked him to or not, and ended up having quite a heated debate with one signalman who was most insistent that he didn’t need to call back and he’d much prefer that he didn’t call back. Quite reasonably, he asked how the signalman would know that it was safe to run trains if he couldn’t be certain that he’d crossed safely.

Thats correct, assume crossing clear and don’t replace signals if under 3 minutes is the national SOP.
Our rule was a link only instruction, then box instruction after the north wales incident.

I always replaced and collared, and maintained until they called back with crossing clear.
Safety, safety, safety....until it starts delaying trains aka costing money.....not on my panel.

Is the three minutes an official thing now? It was always up for debate. Personally I’d provide signal protection for anything bigger than a car or van without trailer. I remember someone coming up with three minutes as the threshold, but then someone with an obviously large piece of machinery could quite reasonably state that they only need two minutes!

 

[Tom Quinne]

Three minutes, no reversion of signals - unless it’s a long/slow or low vehicle etc.
Nationally it still stands, but on my route it’s signals ON and collared, the caller must also be asked to call back crossing clear.

As an aside, we also have to apply additional protection to ALL line blocks, where as other routes don’t.

 

[Tom Quinne]

Even before the call back was required, I insisted in it. Im of the same view as your farmer chap, how do I know he’s not fallen over broke it ankle, his mobile has flew out of his hand out of reach with him no trapped unable to clear in the four foot of the Down line with a train approaching because the signaller has (according to the NOP) assumed hes cleared ?

Im naturally cynical of human kind, and don’t trust MOPs to ensure I don’t end up gripping a brass rail, hence I take all reasonable steps to ensure I remain in control of the interface between me and them.

 

[Taunton]

how do I know he’s not fallen over broke it ankle, his mobile has flew out of his hand out of reach with him no trapped unable to clear in the four foot of the Down line with a train approaching.

Actually far more realistic than these visions like something out of a hackneyed scene from a Hollywood scene in Lassie (no offence, sir!) is that the vehicle has stranded on the crossing. A significant number of user-worked crossings rise from field level either side to rail level, a heavily laden farm vehicle clearing the field is liable to strand going over the convex aspect of the crossing - in fact, I believe there was a serious accident at a Netherlands UWC just a couple of weeks ago precisely in this situation. The old road crossing signage I recall used to have a pictogram sign showing exactly this for an HGV low loader, with a plate saying "Risk of Grounding".

 

[ForTheLoveOf]

Actually far more realistic than these visions like something out of a hackneyed scene from a Hollywood scene in Lassie (no offence, sir!) is that the vehicle has stranded on the crossing. A significant number of user-worked crossings rise from field level either side to rail level, a heavily laden farm vehicle clearing the field is liable to strand going over the convex aspect of the crossing - in fact, I believe there was a serious accident at a Netherlands UWC just a couple of weeks ago precisely in this situation. The old road crossing signage I recall used to have a pictogram sign showing exactly this for an HGV low loader, with a plate saying "Risk of Grounding".

It strikes me that, without any information as to the exact gradients, embankment elevations etc. involved, just warning about a "risk of grounding" is a bit pointless. What are you supposed to do if you're unsure whether you can fit over? Of course people will say "don't cross", well what do you do if the only way in and out of the field is via this UWC? Might be OK for farmers that are familiar with the limits of their local UWCs but there is a lot of scope for error there surely.

 

[Tom Quinne]

Actually far more realistic than these visions like something out of a hackneyed scene from a Hollywood scene in Lassie (no offence, sir!) is that the vehicle has stranded on the crossing. A significant number of user-worked crossings rise from field level either side to rail level, a heavily laden farm vehicle clearing the field is liable to strand going over the convex aspect of the crossing - in fact, I believe there was a serious accident at a Netherlands UWC just a couple of weeks ago precisely in this situation. The old road crossing signage I recall used to have a pictogram sign showing exactly this for an HGV low loader, with a plate saying "Risk of Grounding".

My example was prior to him opening the far side gate, getting back into his vehicle and then that suffering a failure and becoming stranded on the crossing lol

Either way the “assume safe, no call back” method is potentially fatally flawed.

 

[CrispyUK]

Even before the call back was required, I insisted in it. Im of the same view as your farmer chap, how do I know he’s not fallen over broke it ankle, his mobile has flew out of his hand out of reach with him no trapped unable to clear in the four foot of the Down line with a train approaching because the signaller has (according to the NOP) assumed hes cleared ?

Im naturally cynical of human kind, and don’t trust MOPs to ensure I don’t end up gripping a brass rail, hence I take all reasonable steps to ensure I remain in control of the interface between me and them.

What happens if a user doesn’t bother to call back to confirm the crossing is clear despite your request they do so? I am guessing you would need to contact the driver of the next train and instruct them to proceed at caution, slowly enough that they can stop short of the crossing if it is obstructed once they have a visual on it? But could be completely wrong.

Do all crossings have a phone on each side of the line? The signage shown in an earlier post suggests there is no need to make contact if crossing the line on foot, but such a user could equally end up tripping and getting stuck on the line infront of an oncoming train. Appreciate you can only protect the crossing for users you know are there, but from the signs I wouldn’t know if it was “allowed” or appropriate to call if I wasn’t crossing with vehicles or animals.

 

[Spartacus]

What happens if a user doesn’t bother to call back to confirm the crossing is clear despite your request they do so? I am guessing you would need to contact the driver of the next train and instruct them to proceed at caution, slowly enough that they can stop short of the crossing if it is obstructed once they have a visual on it? But could be completely wrong.

Do all crossings have a phone on each side of the line? The signage shown in an earlier post suggests there is no need to make contact if crossing the line on foot, but such a user could equally end up tripping and getting stuck on the line infront of an oncoming train. Appreciate you can only protect the crossing for users you know are there, but from the signs I wouldn’t know if it was “allowed” or appropriate to call if I wasn’t crossing with vehicles or animals.

I’d have thought the vast majority of people should be able to roll or drag themselves out of the way if they’ve fallen if they’ve given themselves anything even approaching appropriate crossing time before a train, and if they were frail enough not to be able to do so themselves I’d be surprised if they were out alone.

 

[Llanigraham]

What happens if a user doesn’t bother to call back to confirm the crossing is clear despite your request they do so? I am guessing you would need to contact the driver of the next train and instruct them to proceed at caution, slowly enough that they can stop short of the crossing if it is obstructed once they have a visual on it? But could be completely wrong.

Quite easy.
You caution the next train into the section, passing the section signal at danger, and tell them where the "incident" is so they approach with caution being prepared to stop, and then to report back.

Do all crossings have a phone on each side of the line? The signage shown in an earlier post suggests there is no need to make contact if crossing the line on foot, but such a user could equally end up tripping and getting stuck on the line infront of an oncoming train. Appreciate you can only protect the crossing for users you know are there, but from the signs I wouldn’t know if it was “allowed” or appropriate to call if I wasn’t crossing with vehicles or animals.

Depends.
All 3 of the crossings we controlled were phone equipped, 2 farm crossings,1 bridleway, but I know of numerous footpath crossing that have no phone.

 

[Tomnick]

Three minutes, no reversion of signals - unless it’s a long/slow or low vehicle etc.
Nationally it still stands, but on my route it’s signals ON and collared, the caller must also be asked to call back crossing clear.

As an aside, we also have to apply additional protection to ALL line blocks, where as other routes don’t.

Where's that (the three minutes) actually published, though? Was it ever formalised in proper company instructions? The signalling regs still don't specify what comes under the definition of "large, low or small moving vehicles".

Additional protection for line blockages - doesn't that just drag out work being done between trains, e.g. patrolling, and mean that they end up taking more line blocks and thus more risk overall?

Even before the call back was required, I insisted in it. Im of the same view as your farmer chap, how do I know he’s not fallen over broke it ankle, his mobile has flew out of his hand out of reach with him no trapped unable to clear in the four foot of the Down line with a train approaching because the signaller has (according to the NOP) assumed hes cleared ?

Im naturally cynical of human kind, and don’t trust MOPs to ensure I don’t end up gripping a brass rail, hence I take all reasonable steps to ensure I remain in control of the interface between me and them.

I generally did insist on a call back and provide signal protection. Thankfully my position was never really put to the test but I think it likely that they'd have tried to stick the delay to me if I'd had to caution because a user crossing with just a car had failed to call back! Sometimes I had to take a more practical view, e.g. the busy UWC not far outside the home signal at one box, where you'd typically have accepted a train ten minutes before it actually reached the crossing. How many times is the under-pressure DPD driver, a regular visitor, going to sit and wait for a train that's ten minutes away before deciding "f**k it, I'm just going to take a chance this time"? It's pretty unlikely, really, that someone crossing a well maintained crossing in a small vehicle is going to get stuck (not really much more so than at an AHB?), but if users get fed up and start routinely crossing without phoning, it's far more likely that there'll be an accident.

 

[Tom Quinne]

Where's that (the three minutes) actually published, though? Was it ever formalised in proper company instructions? The signalling regs still don't specify what comes under the definition of "large, low or small moving vehicles".

Additional protection for line blockages - doesn't that just drag out work being done between trains, e.g. patrolling, and mean that they end up taking more line blocks and thus more risk overall?

I generally did insist on a call back and provide signal protection. Thankfully my position was never really put to the test but I think it likely that they'd have tried to stick the delay to me if I'd had to caution because a user crossing with just a car had failed to call back! Sometimes I had to take a more practical view, e.g. the busy UWC not far outside the home signal at one box, where you'd typically have accepted a train ten minutes before it actually reached the crossing. How many times is the under-pressure DPD driver, a regular visitor, going to sit and wait for a train that's ten minutes away before deciding "f**k it, I'm just going to take a chance this time"? It's pretty unlikely, really, that someone crossing a well maintained crossing in a small vehicle is going to get stuck (not really much more so than at an AHB?), but if users get fed up and start routinely crossing without phoning, it's far more likely that there'll be an accident.

I’ve no idea if or where the “3 minute rule” was, or if it was even a rule.

Regards additional protection, yes! It does increase time going on, and off.

We’re all axle counter so it’s EPR on regardless of whether their planned or required.

Patrollers needing to Cross a 10 meter long bridge need a line block and EPR now.

I understand there is a heavy push to zero or making it as difficult as possible to have red zone works, so pretty much all work is in green zone line blocks - which increase our workload.

 

[Tom Quinne]

What happens if a user doesn’t bother to call back to confirm the crossing is clear despite your request they do so? I am guessing you would need to contact the driver of the next train and instruct them to proceed at caution, slowly enough that they can stop short of the crossing if it is obstructed once they have a visual on it? But could be completely wrong.

Do all crossings have a phone on each side of the line? The signage shown in an earlier post suggests there is no need to make contact if crossing the line on foot, but such a user could equally end up tripping and getting stuck on the line infront of an oncoming train. Appreciate you can only protect the crossing for users you know are there, but from the signs I wouldn’t know if it was “allowed” or appropriate to call if I wasn’t crossing with vehicles or animals.

Yes, no call back you caution the next train over to ensure it’s clear.

All (in my area) UWC have phones both sides, our level crossing managers are very hot on ensuring phones are working correctly.

 

[elbows47]

A slight sidetrack to this discussion - does anyone know which unit was involved, and was the driver OK?