How quick is ETCS / ERTMS when switching at junctions?

[Nottingham59]

Any ETCS experts out there?

I understand that ETCS allows more train paths through complex junctions. I'm trying to understand just how quick the system is.

For example, when train A passes through a junction followed by train B, then I assume the following steps are needed:

• ETCS detects that train A has cleared the point of conflict
• The signalling system switches the points (assuming the route has previously been set up by the signaller)
• ETCS detects that the points have switched and proves they are locked
• ETCS issues movement authority to train B
• Train B receives movement authority
• B calculates a revised braking curve and a new maximum speed
• B displays the new data to the driver
• Driver responds (or is the speed controlled automatically?)

Presumably this whole process takes just a few seconds. If I was trying to model it, how long should I allow?

Thanks in advance.

(If it helps, the specific situation is trying to understand how many trains per hour could traverse the throat of Old Oak Common acting as a terminus for HS2.)

 

[IanXC]

My understanding is that ETCS will ask the interlocking for the route the next train wants, so in many ways the response time is likely to be similar to conventional signalling. In the ECML scheme ETCS is the route to providing the information, but the interlockings remain as they currently are - whether the HS2 interlockings will be materially different I'm not sure. The main difference relevant to your question as I understand it, is that on approach to the junction ETCS will give train B a speed profile that optimises the overall time through the junction - avoiding coming to a stand and then having to move off again.

 

[edwin_m]

I presume you are trying to calculate the minimum spacing between these trains, which happens when train B gets an extended movement authority (MA) just before it was about to brake for the end of its previous MA. So you have to allow for a braking reaction and application time as well, as although it isn't actually invoked the system has to allow for that possibility if the new MA isn't received in time.

There was a simple analysis by Andrew McNaughton on the HS2 website at one time, but it no longer appears to be available. Based on a quick search section 6 of https://www.whatdotheyknow.com/requ...Capacity Technical Note Final 1.1 Private.pdf
may be of interest.

In this section we calculate needed headway in various situations (open line running, at a
diverging turnout, with slowdowns on the line, etc). We also review the approach to headway
estimation in various situations that has been taken by HS2 in the document Signalling
Headways and Maximal Operational Capacity on High Speed Two London to West Midlands
Route, version 3, dated 1 August, 2011 [21].

 

[Nottingham59]

in many ways the response time is likely to be similar to conventional signalling.

Thank you. So what would be a reasonable guess as to what that time will be?

From the point that A clears the junction to the driver getting a proceed aspect would be just a few seconds? Plus maybe 5-10 seconds to respond if the train is at a stand, and the driver is expecting a green light imminently? Does ETCS give an audible notification if a standing train has authority to move? Is 30 seconds in total more than enough?

There was a simple analysis by Andrew McNaughton on the HS2 website at one time, but it no longer appears to be available. Based on a quick search section 6 of https://www.whatdotheyknow.com/requ...Capacity Technical Note Final 1.1 Private.pdf
may be of interest.

Thanks. Page 39 seems to have the data I need.

EDIT: From what I can see, the timings are:

Turnout operation time 9
Train detection (TD) system reporting delay time 5
Movement authority (MA) update transmission time 2
Train On Board ETCS reaction time 1
Worst case driver response time 8
Automatic Train Operation (ATO) response time 0

So 25 seconds in total

 

[Harpo]

Timetablers used to be allowed to plan a sub-standard headway through a junction by using ‘SOY’, starting a second train off on 1Y, so maybe 1 -1.5 min apart rather than 2-3.

The first & faster train would move away, leaving the follower running on 2Y then G at the standard headway. Forcing the second train to wait for a green puts the two trains 1.5 headways or more apart.

ETCS potentially allows stuff like this, which was eliminated, to be re-introduced.

 

[MarkyT]

Timetablers used to be allowed to plan a sub-standard headway through a junction by using ‘SOY’, starting a second train off on 1Y, so maybe 1 -1.5 min apart rather than 2-3.

The first & faster train would move away, leaving the follower running on 2Y then G at the standard headway. Forcing the second train to wait for a green puts the two trains 1.5 headways or more apart.

ISTR, after all the worry about ding ding and away incidents, someone raised a safety concern about dispatching on yellow. Of course it happened anyway if the first train was running a little late. In the signal drawing office at the time, we all thought it a daft policy, wasting headway. The trains gradually space themselves out as the front one races away while the rear one struggles into motion. Here's an illustration:

Dawlish Warren Up Loop Passing Manoeuvre - 18th October 2014

A First Great Western HST overtakes a Sprinter at Dawlish Warren.

ETCS potentially allows stuff like this, which was eliminated, to be re-introduced.

Agreed. SOY should once again be allowed with colour lights too.

 

[dodecahedron67]

Idk about other TOCs but we are absolutely allowed to dispatch against single yellows at mine, with one exception - York P4 Up direction

Indeed there are some places where you can see the signal, but it isn’t classed as a ‘starting signal’ (AWS ramp not within the platform & train can fit between platform and signal) so we can dispatch even if it’s red, Newton-le-Willows on the Up Chat Moss, Normanton Down L&Y towards Leeds/Castleford, for example

 

[JN114]

ETCS doesn’t detect trains - at least in any “Level” currently implemented.

It is Movement Authorities sent from the signalling system via the Radio Block Centre and presented to the driver (or ATO system) on an in cab display as a distance-to-go and calculated target speed.

Train detection and route setting is still as it is in ordinary signalling.

There is probably a minor delay for transmission/translation/calculation from Signalling System > Radio Block Centre > Cab Display vs legacy signalling; but probably measured only in 10s of milliseconds from practical examples I’ve seen.

 

[plugwash]

With traditional four aspect signals, blocks must be long enough so that a train at line speed can stop within two and a bit blocks. Lineside signals must be physically placed where the driver can see them and where there is something to mount them. On multi-track railways signals on all the tracks are normally placed in the same locations to reduce the chance of mis-reads.

My understanding is that ETCS level 2 lifts those constraints which allows in-principle, for higher capacity designs with much smaller blocks. Of course there is a cost to installing such a design............

 

[The Planner]

Timetablers used to be allowed to plan a sub-standard headway through a junction by using ‘SOY’, starting a second train off on 1Y, so maybe 1 -1.5 min apart rather than 2-3.

The first & faster train would move away, leaving the follower running on 2Y then G at the standard headway. Forcing the second train to wait for a green puts the two trains 1.5 headways or more apart.

ETCS potentially allows stuff like this, which was eliminated, to be re-introduced.

Not aware of anywhere where we do that. You plan on greens.

 

[Bald Rick]

Does ETCS give an audible notification if a standing train has authority to move?

There is no difference inwhat the signaller (or Automatic Routing system) does.

There are two differences though:

1) ETCS cycle time. This is how frequently the train receves contact from the Radio Block Centre. IIRC this can vary but is typically 5 seconds. Depending when the signalling system has set the route and confirmed clear to the next marker board, this means the train will receive the movement authority in the range of 0-5 seconds later.

2) The movement authority is to a specific block marker, and not necessarily the same distance as a conventional signalling ‘block’. Therefore if the distance clear is shorter than a conventional signalling block length, the driver can receive the MA earlier than under conventional signalling, and also (of course) receive the MA at any time, not just when they can observe an approaching signal. Yes there is an audible double chime to advise, not unlike the original SMS chime 30 years ago.

So, the activities between train A moving and points being swung are exactly the same as under conventional computer based signalling (same interlocking, same route setting procedure). What happens after that is a bit quicker but only if train B is in the braking curve on the approach to the formerly closed block marker, in a position where the driver would have been unable to see the next signal; and the diverging route ahead the train is routed to is clear to a distance that enables a speed higher than it would be had it been routed on the non-diverging route.

It certainly helps when trains are running at headways that require lower speeds than the linespeed, as trains can be closer together and kept moving (at slower speeds).

Welwyn viaduct will be a good test for this when the trains start using the commissioned ETCS.


Quite tricky to explain in words!

 

[Harpo]

Quite tricky to explain in words!

Very! A second subsequent train over a junction has to be at braking distance and a bit in rear of the junction while it’s fouled, and that distance will vary by train speed. ETCS won’t change that.

 

[edwin_m]

Very! A second subsequent train over a junction has to be at braking distance and a bit in rear of the junction while it’s fouled, and that distance will vary by train speed. ETCS won’t change that.

With conventional signalling the headway is based on always approaching green signals (excepting approach control, but let's not go there), so train separation is nominally independent of train speed. With ETCS a slower train or one with better braking can approach closer to the train in front before it has to slow down, although for Level 2 this is subject to train detection sections being suitably short and in the right places.

There's also the complication around junctions that the train may be using a lower deceleration rate to slow down for the speed restriction (for example using regenerative braking only) but might increase its deceleration (apply a friction brake) if needing to slow down for a conflicting move. Basing headways on the greater deceleration allows them to be shorter while maximising energy saving, as most of the time the conflict will clear before the friction brake is needed especially if there is a driver advisory system or the ATO equivalent suggesting an optimum speed profile. All of which complicates the calculations even more...

 

[IanXC]

Idk about other TOCs but we are absolutely allowed to dispatch against single yellows at mine, with one exception - York P4 Up direction

Indeed there are some places where you can see the signal, but it isn’t classed as a ‘starting signal’ (AWS ramp not within the platform & train can fit between platform and signal) so we can dispatch even if it’s red, Newton-le-Willows on the Up Chat Moss, Normanton Down L&Y towards Leeds/Castleford, for example

(Presumably not dispatching against a single yellow also applies to the mid platform signals at Leeds and Newcastle)

 

[dodecahedron67]

(Presumably not dispatching against a single yellow also applies to the mid platform signals at Leeds and Newcastle)

On the rare occaision we stop in a position to use these, we are allowed to dispatch against SYs at Leeds (to my knowledge), Newcastle I’m not sure as I’ve never had to do it!

 

[Nottingham59]

There's also the complication around junctions that the train may be using a lower deceleration rate to slow down for the speed restriction (for example using regenerative braking only) but might increase its deceleration (apply a friction brake) if needing to slow down for a conflicting move. Basing headways on the greater deceleration allows them to be shorter while maximising energy saving, as most of the time the conflict will clear before the friction brake is needed especially if there is a driver advisory system or the ATO equivalent suggesting an optimum speed profile. All of which complicates the calculations even more...

Thanks. I'm coming to understand that for HS2 trains approaching OOC!

[Off topic: I'm assuming Trains will brake at less than Normal Service Brake (which allows regeneration) to halt at the end of the platforms. But if the Victoria Crossing Box (1.2km from the buffers) is set against them, then the trains need to apply Full Service Brake from a point 4km short of the buffers in order to stop in time. So the VCB switches need setting a full 120 seconds before the train reaches them, if they are not to interrupt smooth operations.

Northbound, trains launch from stationary, and the VCB switch can be set after departure.]

Quite tricky to explain in words!

I understand. Thanks for the explanation.

 

[IanXC]

On the rare occaision we stop in a position to use these, we are allowed to dispatch against SYs at Leeds (to my knowledge), Newcastle I’m not sure as I’ve never had to do it!

Interesting. Other TOCs consider dispatch against a mid platform single yellow (excluding towards the through line) an operating incident! If it were to happen and the train in question came to a stand at the platform end signal, they'd dispatch it again.

 

[MarkyT]

With conventional signalling the headway is based on always approaching green signals (excepting approach control, but let's not go there), so train separation is nominally independent of train speed. With ETCS a slower train or one with better braking can approach closer to the train in fr

Minimum colour light signal spacing is based on worst-case braking. You can't exceed braking distance significantly from first caution to red either. Spacing over 150% SBD (standard braking distance) isn't allowed normally, thus signalling headway is 'tuned' to a particular speed. Excessive difference in speed between subsequent trains will have complex effects on deliverable headway. A significant temporary speed restriction for all traffic can also have a devastating effect on throughput. A short 20mph engineering slack in a section tuned for dense 80-90mph traffic could easily double section time for the block concerned. If there were trains approaching at maximum technical frequency at normal line speed, they simply couldn't be accommodated through the restricted block and each subsequent train would not only pick up the extra journey time of the slow down but would be held increasingly longer for block clearance. Luckily, planning headway is typically substantially more than technical headway, so this can be managed to an extent. ETCS with significantly shorter blocks through very busy areas should be able to handle such a scenario much better, with slower following trains able to bunch up more closely through any slow zones.

 

[edwin_m]

[Off topic: I'm assuming Trains will brake at less than Normal Service Brake (which allows regeneration) to halt at the end of the platforms. But if the Victoria Crossing Box (1.2km from the buffers) is set against them, then the trains need to apply Full Service Brake from a point 4km short of the buffers in order to stop in time. So the VCB switches need setting a full 120 seconds before the train reaches them, if they are not to interrupt smooth operations.

Northbound, trains launch from stationary, and the VCB switch can be set after departure.]

Probably not off-topic at all. There is a place where the regenerative braking curve to stop at the buffer stops intersects the steeper full braking curve to stop short of the crossing box. To run without being checked, the train needs to have its movement authority all the way to the buffer stops in enough time before reaching that position that the brake application doesn't need to be initiated. Not sure where your 4km comes from but if it's the braking distance from line speed then it may be too far - you need the distance to where the curves intersect.

 

[Nottingham59]

Probably not off-topic at all. There is a place where the regenerative braking curve to stop at the buffer stops intersects the steeper full braking curve to stop short of the crossing box. To run without being checked, the train needs to have its movement authority all the way to the buffer stops in enough time before reaching that position that the brake application doesn't need to be initiated.

Not sure where your 4km comes from but if it's the braking distance from line speed then it may be too far - you need the distance to where the curves intersect.

Braking distance from HS2 line speed is 12km under Normal Service Braking. (I would expect braking to be better than that in the tunnel, due to aerodynamic effects, but I haven't modelled these.)

The 4km is the point where the Normal Service Braking curve to stop at the platform crosses the Full Service Braking curve to stop before the Victoria Crossing Box. At a point where the train is still doing 230km/h.

 

[Bald Rick]

With conventional signalling the headway is based on always approaching green signals (excepting approach control, but let's not go there), so train separation is nominally independent of train speed. With ETCS a slower train or one with better braking can approach closer to the train in front before it has to slow down, although for Level 2 this is subject to train detection sections being suitably short and in the right places.

This os where it hets interesting, as planning headway is based on approaching green signals with the preceeding train at the prevailing linespeed (with some adjustments for freight etc). There is no standard method of plannign services at different headway for different speeds.

What ETCS will do with shorter sections will be to enable more rapid platform reoccupation (as seen in the TL core) and more rapid following of services at converging junctions (as seen at Welwyn viaduct), albeit this latter facility will be primarily to reduce delay, rather than fit in more services.

 

[MarkyT]

Here's a video I made to illustrate the speed and headway problem using Transport Fever 2.
I made a circular 25km 220kph railway with signal blocks ~1000m long. I set Twelve 4-car bullet trains travelling endlessly at full speed spaced around 3 blocks apart. Effectively, the line is operating at maximum technical headway. Note I had to place two stations on the far side of the circle as the game wouldn't create a line without them. These have closing up arrangements on approach so some of the trains get much closer to each other on that side. That's why 25km/12 trains doesn't correspond precisely with the observed free-running spacing.

After introducing a 60kph speed restriction for the extent of one block I saw traffic backing up very quickly due to the extended time required for each train to traverse the speed-limited fixed-length block section. Trains were stopping or nearly coming to a stand at multiple signals on approach, severely constraining throughput. Once back up to speed after leaving the restriction the trains were spaced around 5 block sections apart with no way to get any closer together, until they got round the circle and joined the back of the queue again for the restriction.

Then I added an extra signal midway through the speed limited block, dividing it into two sections. While trains must still slow down through the restriction, the extent of backing up on approach is mitigated. Trains are no longer coming to a stand.

Fortunately the game's signalling functionality is rather like ETCS so you can have blocks as short as you like and the trains automatically work out their own braking envelopes regardless of how many sections are invloved.