Why can't level crossing activation points be set according to speed

[Edders23]

the other day I got caught at Tallington crossing (no surprise there) but from barriers down to the arrival of the class 66 trundling Sibelco minerals past at 20MPH was approximately 5 and half minutes

The barriers generally are activated when a fast passenger is about 2 and half minutes away which is often 3 to 3 and a half minutes given that very few trains are doing anything near 125 at that point

I can understand stopping distance for a class91, Hst or azuma at full line speed but does a shed travelling at 20MPH with a dozen empty wagons need the barriers to go down when it is that far away ? or a 158 at 75mph surely they can activate the crossing a bit later without compromising safety

 

[MadMac]

For automatic or “auto-lower” crossings, the problem is that nobody’s yet come up with a US-style “predictive” or “constant warning time” system that’ll work on an electrified line. There is, IIRC, one such system in Norfolk on a non-electrified line, but the supplier had a lot of trouble getting it approved by Railtrack.

There used to be a concept called “Speed Discrimination” that was used at some older AHB installations, but there’s so much of a mix of traffic types that it can’t be “all things to all men”.

 

[NSEFAN]

the other day I got caught at Tallington crossing (no surprise there) but from barriers down to the arrival of the class 66 trundling Sibelco minerals past at 20MPH was approximately 5 and half minutes

The barriers generally are activated when a fast passenger is about 2 and half minutes away which is often 3 to 3 and a half minutes given that very few trains are doing anything near 125 at that point

I can understand stopping distance for a class91, Hst or azuma at full line speed but does a shed travelling at 20MPH with a dozen empty wagons need the barriers to go down when it is that far away ? or a 158 at 75mph surely they can activate the crossing a bit later without compromising safety

If its a fully controlled crossing, the activation would be based on the position of signals, and essentially making sure that the driver sees only green signals as they approach. Whilst a slower freight can be activated a bit later, it wouldn't make that much difference.

For an automatic crossing, I would imagine that it's easier from a safety perspective to have a simpler system with a fixed strike-in point. What benefit is there to the railway in reducing crossing times, and would it outweigh the cost of making sure the time-predictive system is safe?

I believe that there are certain locations where the signaller must tell the signalling if a train is fast or stopping, which alters how the signals interact with an automatic crossing. However this is a fairly simple and the "stopping" mode sets the protecting signal to danger as a means of maintaining safety.

 

[edwin_m]

For an automatic crossing, I would imagine that it's easier from a safety perspective to have a simpler system with a fixed strike-in point. What benefit is there to the railway in reducing crossing times, and would it outweigh the cost of making sure the time-predictive system is safe?

An automatic half barrier carries the risk of road vehicles zigzagging round the barriers, which becomes much more likely if the crossing time is longer than what their drivers consider to be reasonable. I don't know the details but I think this is covered somewhere in the crossing risk assessment process.

I believe the US crossing predictor system mentioned above sends an electrical signal along the rails, which is reflected back by the train wheels and gives a continuous indication of where the train is. It's not really surprising that this sort of system is affected by traction return current, which can put all sorts of voltages and frequencies into the rails. It's possible to imagine a system based on measuring the time between passing a pair of sensors, but it needs to take account of the possibility that the train is accelerating - and something like a light engine can accelerate a lot! So I think it would either need a huge number of sensors or would have to build in margins of safety that made it not much better than what is provided now.

 

[John Webb]

Would the 'Digital Railway' (if I correctly understood what I've read about it) where trains will be located by GPS give the facility to tailor crossing closing times to the approaching train speed?

 

[Highlandspring]

For automatic or “auto-lower” crossings, the problem is that nobody’s yet come up with a US-style “predictive” or “constant warning time” system that’ll work on an electrified line. There is, IIRC, one such system in Norfolk on a non-electrified line, but the supplier had a lot of trouble getting it approved by Railtrack.

Haugh of Tullymet MSL user worked crossing on the Highland Main Line also uses the predictive system you mention.

I’ll need to dig out the scheme plan from the 2013 MCB-OD upgrade at Kirknewton because something in the back of my mind tells me there was some funny arrangement with the train describer to try and get a consistent strike in time...

 

[Ianno87]

Would the 'Digital Railway' (if I correctly understood what I've read about it) where trains will be located by GPS give the facility to tailor crossing closing times to the approaching train speed?

In theory, ETCS Level 3 (which has no practical applications as we speak) should calculate a continuously changing braking curve in front of every train, depending on its speed and braking profile. So, in theory, some whizzy technology could start the level crossing sequence in time to have it down and cleared just in time for the 'front' of the braking curve reaching the crossing (or a 'safe' point just before).

But will probably only be a minor dent in overall down times overall; not going to make rip-roaring difference.

 

[edwin_m]

Would the 'Digital Railway' (if I correctly understood what I've read about it) where trains will be located by GPS give the facility to tailor crossing closing times to the approaching train speed?

GPS on its own is not reliable enough to be used in a safety-critical application such as ensuring that level crossing barriers close at the right time. And the architecture of such a system is also challenging - how does the level crossing know which train is approaching so as to interrogate its GPS, without relying on things like the train describer and whatever system records the allocation of units to headcodes, neither of which is safety-critical either?

In theory, ETCS Level 3 (which has no practical applications as we speak) should calculate a continuously changing braking curve in front of every train, depending on its speed and braking profile. So, in theory, some whizzy technology could start the level crossing sequence in time to have it down and cleared just in time for the 'front' of the braking curve reaching the crossing (or a 'safe' point just before).

But will probably only be a minor dent in overall down times overall; not going to make rip-roaring difference.

I'd like to clarify that by pointing out that ETCS level 3 doesn't exist yet - there might be quite a lot of routes wanting to apply it if it did.

Actually I think Level 3 would work quite well with crossing prediction, especially for the MCB-OD type crossings that are replacing many AHBs but have longer road closure times because the railway normally aims to close the crossing in time for the train to run through on green signals. Under Level 3 the train would have a "movement authority" instead, which reserves a section of track ahead of it according to its current speed and braking abilities. A train running more slowly will have a shorter movement authority, although a freight train will have a longer one than a passenger train at the same speed because its deceleration rate is less. So it's "simply" a question of the central computer noticing that the train will soon need a movement authority across the crossing and initiating the closing sequence in enough time. If the crossing didn't operate for some reason (such as detecting an obstacle) the movement authority couldn't be extended and the train would be stopped before the crossing. Some of this may even work with ERTMS level 2.

 

[MarkyT]

Actually I think Level 3 would work quite well with crossing prediction, especially for the MCB-OD type crossings that are replacing many AHBs but have longer road closure times because the railway normally aims to close the crossing in time for the train to run through on green signals. Under Level 3 the train would have a "movement authority" instead, which reserves a section of track ahead of it according to its current speed and braking abilities. A train running more slowly will have a shorter movement authority, although a freight train will have a longer one than a passenger train at the same speed because its deceleration rate is less. So it's "simply" a question of the central computer noticing that the train will soon need a movement authority across the crossing and initiating the closing sequence in enough time. If the crossing didn't operate for some reason (such as detecting an obstacle) the movement authority couldn't be extended and the train would be stopped before the crossing. Some of this may even work with ERTMS level 2.

Good thinking there. I expect L2 already incorporates a mechanism whereby a train is continually reporting in its position and speed, so ARS can set movement authority forward as required. For this purpose the details will be those of the front active cab. Where such positioning is required for block clearance, the train also needs to know about its length and continuity in order to report its rear position, but for crossing strike in only front of train information is relevant. For strike out a small length of fixed train detection can be retained over the crossing along with treadles and the OD system to prove the roadway is clear again after passage.

 

[Edders23]

Some of the replies mention automated crossings as far as I am aware Tallington is manually operated from Peterborough power signal box which is partly why I asked the question

it has been known for the barriers to stay down for 10 minutes after a train and misteriously then go up after the signalmen realise their mistake !!!!

 

[edwin_m]

Good thinking there. I expect L2 already incorporates a mechanism whereby a train is continually reporting in its position and speed, so ARS can set movement authority forward as required. For this purpose the details will be those of the front active cab. Where such positioning is required for block clearance, the train also needs to know about its length and continuity in order to report its rear position, but for crossing strike in only front of train information is relevant. For strike out a small length of fixed train detection can be retained over the crossing along with treadles and the OD system to prove the roadway is clear again after passage.

In L2 the train is definitely detected by conventional means. I'm pretty sure it is then allocated a movement authority of a certain length and regulates its own speed to stop before the end of the movement authority. I don't think the interlocking knows the speed or braking capability of the train so it would just allocate the same movement authority to all trains, equivalent to the braking distance from line speed.

It would therefore be possible in principle for a non-vital system to deduce the speed and/or braking capability of the train, and tell the vital interlocking to allocate certain trains a shorter movement authority on the approach to the crossing on the basis they are probably going more slowly. If the non-vital system gets this wrong then the train gets a shorter movement authority than it needs to continue at current speed and has to slow down - operationally inconvenient but not hazardous.

 

[Legolash2o]

An automatic half barrier carries the risk of road vehicles zigzagging round the barriers, which becomes much more likely if the crossing time is longer than what their drivers consider to be reasonable. I don't know the details but I think this is covered somewhere in the crossing risk assessment process.

I tend to agree with this. Pedestrians and drivers know they could be stuck behind a crossing for a long while and therefore increasing the temptation to risk driving/running through it.

There was once or twice a few years ago, when I got stopped by a crossing walking home. The train took about 5 minutes to eventually pass, people where getting angry asking where is this train?!??! The barriers then remained down as another train was going to set off at the adjacent station (Bridlington). For some reason the barriers remained down even though no train was moving for another few minutes. It took well over 10 minutes for them to rise and the built up traffic to start moving again.

I know it's safety, but you can argue that doing that increases the danger as people will assume they train is miles away and wont be there for a while. I completely understand why people rush, put their foot down, whatever when they see the flashing lights.

 

[Ianno87]

In L2 the train is definitely detected by conventional means. I'm pretty sure it is then allocated a movement authority of a certain length and regulates its own speed to stop before the end of the movement authority. I don't think the interlocking knows the speed or braking capability of the train so it would just allocate the same movement authority to all trains, equivalent to the braking distance from line speed.

AIUI under L2, the *train* knows its braking capability and braking curve, so requests from the interlocking the appropiate level of Movement Authority ahead of itself (as distatated by the fixed block sections available) to always protect this, so the train can run unimpeded (unless there is another train occupying a necessary block section ahead)

 

[John Webb]

Thanks for the replies re the possible use of GPS.

 

[hooverboy]

in the next 5-10 years we are going to have active traffic management systems on the roads,meaning that your car will be in communication with various sensors attatched to crossings and lights,driven by AI(or something similar) which will calculate amount of traffic and speeds etc, to adjust the timings of lights to optimise traffic flow.

the scale of this is enormous compared to what the railways are requiring,and it will be done way before the railways get anything approaching it.
One has to ask why.

speed is not the issue, as there are already functional lte-r systems in korea(designed by nokia siemens networks) that have been tested to 450km/h.
In fact mobile comms is more difficult for urban roads in my view, because you have to throw into the mix a large amount of buildings and structures which can absorb/reflect signals in strange ways depending on their composition.
rail is a great deal more open fields,so the signal is cleaner and more predictable.

what I do see missing from the railway is a known standard of interoperability, such as 3gpp does for mobile networks globally,and to which all equipment providers must comply,and be proven to comply by independent test houses.

 

[RichardGore]

Under any ETCS full supervision the train has to know its own braking capability and braking curve, including L1. In L2 and L3 the train can also tell the RBC its current position (packets 0 and 1) but while this does include speed, it doesn’t include braking curve information, so the RBC can’t in general know the braking characteristics.

 

[Bald Rick]

In L2 the train is definitely detected by conventional means. I'm pretty sure it is then allocated a movement authority of a certain length and regulates its own speed to stop before the end of the movement authority. I don't think the interlocking knows the speed or braking capability of the train so it would just allocate the same movement authority to all trains, equivalent to the braking distance from line speed.

It would therefore be possible in principle for a non-vital system to deduce the speed and/or braking capability of the train, and tell the vital interlocking to allocate certain trains a shorter movement authority on the approach to the crossing on the basis they are probably going more slowly. If the non-vital system gets this wrong then the train gets a shorter movement authority than it needs to continue at current speed and has to slow down - operationally inconvenient but not hazardous.

AIUI under L2, the *train* knows its braking capability and braking curve, so requests from the interlocking the appropiate level of Movement Authority ahead of itself (as distatated by the fixed block sections available) to always protect this, so the train can run unimpeded (unless there is another train occupying a necessary block section ahead)

Correct. The interlocking merely controls the ‘space’ available, the Radio Block Centre communicates this to the train, the train knows exactly where it is through on board odometry and checks with fixed balises, and calculates the speed curve appropriately.

Also, whilst L2 uses conventional methods of train detection, there is no limit on block sizes, and trains are not bound by conventional signalling aspect rules. It is perfectly possible to have lots of very short blocks with high speed limits; rather than today where in 4 aspect signalling a train always has 3 clear blocks and a clear overlap in front to get a green aspect, under ETCS L2 it could be 15 short blocks.

There’s no reason why L2 couldn’t be used to improve LX down times, but as @Ianno87 says, it would be a marginal improvement.

 

[Ianno87]

in the next 5-10 years we are going to have active traffic management systems on the roads,meaning that your car will be in communication with various sensors attatched to crossings and lights,driven by AI(or something similar) which will calculate amount of traffic and speeds etc, to adjust the timings of lights to optimise traffic flow.

the scale of this is enormous compared to what the railways are requiring,and it will be done way before the railways get anything approaching it.
One has to ask why.

.

Because, unlike roads (which are a chaotic, unplanned system), the planned flow of movements on the railways already has an optimsed plan to manage flow - otherwise known as the timetable.

And then when stuff is out of sequence, a 'manual' Traffic Management system to manage - i.e. the signaller.

So railways are (pardon the pun) streets ahead of the road network already.

 

[Bald Rick]

in the next 5-10 years we are going to have active traffic management systems on the roads,meaning that your car will be in communication with various sensors attatched to crossings and lights,driven by AI(or something similar) which will calculate amount of traffic and speeds etc, to adjust the timings of lights to optimise traffic flow.

That already happens now, albeit without car - road communication. The traffic lights near my house adjust such that a train peak times, it will show a yellow light when there is a gap in the queue of traffic (subject to a maximum green phase), and at quiet times go green immediately on approach. This isn’t new.

What Active traffic management could do in the coming years is take sat nav data to a central processor and move people’s routing around mid journey to balance out flows and avoid traffic hotspots etc.

 

[Tomnick]

Some of the replies mention automated crossings as far as I am aware Tallington is manually operated from Peterborough power signal box which is partly why I asked the question

it has been known for the barriers to stay down for 10 minutes after a train and misteriously then go up after the signalmen realise their mistake !!!!

It isn’t - it’s worked from Helpston, who control slots on Peterborough’s signals. In normal operation, the barriers should raise automatically after the passage of a train (unless there’s another one signalled across).

 

[colchesterken]

Anyone who has read the posts on level crossings will have seen my winge. The level crossing at East Gate Colchester was upgraded?? a few yrs ago by barrier crossing replacing a man in a signal box overlooking the junction operating the gates
We are trapped in our street by traffic jams caused by the gates being closed for a long time. Speed limit for fast trains 45mph for trains to town 10mph.. I complained to Network Rail then to the M P we had a meeting, I suggested reducing the speed to say 30mph and closing the gap for shutting the gates . We had a tour of Colchester North box and found out the gates close is triggered at the train leaves Colchester North, Town or Wivenhoe It seemed to me the gape could be closed due to low speed limit. The bloke in the orange vest said the expense could not be justified. Just let the whole of the East side of town clog up--not to mention 10000new houses comming between us and Wivenhoe........rant over