A case of poor regulation?

Bikeman78 · 21 Jul 2019

Highlandspring said:
The Cardiff Main Line, Valleys and Vale of Glamorgan workstations are GETS MCS and the other workstations are WestCADs so they'll be using Hitachi's TREsa SARS (Signaller's Assistant Route Setting) rather than Resonate's ARS. The latter system is not too bad but the former makes some real howlers. Or at least it does in the locations where I'm familiar with its use.

Some examples of crazy things ARS does that a signaller would not. Below is a snapshot of trains at Cardiff Queen Street. Let's say 2F31 is six minutes late. It will roll into platform 3 heading south as 2D30 is waiting to depart platform 4 heading north. 2Y37 will be close behind 2F31 but cannot get across Queen St North Jn because the latter is blocking the platform. ARS won't let 2D30 go, it will hold it on a red until until 2F31 departs and 2Y37 arrives, probably putting four minutes delay into 2D30.

http://www.realtimetrains.co.uk/search/advanced/CDQ/2019/07/22/1205-1230?stp=WVS&show=all&order=wtt

Another example, if 2A24 is running late and departs Cardiff Central behind 2D30, there is every chance that ARS will hold 2D30 in platform 4 and route 2A24 around it via platform 5. The two trains take different routes from Queen St so all this does is put four minutes into 2D30 with no benefit to 2A24.

On the main line there are several routes that a passenger train from Newport can take to access platform 3 or 4 at Cardiff Central. If one train stops, e.g. it's booked platform is occupied, ARS will make no attempt to route subsequent trains around even if their booked platform is clear and there is a clear route available.

RailUK Forums

vlad · 21 Jul 2019

JN114 said:
At the end of the day there’s always a winner and loser out of a regulation decision. You won 2 of them:- vs the on time Cardiff and the on time Crewe (the former of which would have lost 10+ minutes waiting for you)

Having checked RTT I'm amazed they let the Crewe train run its full route, as they tend to curtail it at Stoke if it's running late.

Even so, it arrived at Crewe after it was scheduled to depart (it's got a pretty tight turnaround) and didn't make up its delay completely till it got back to Derby.

Tom Quinne · 21 Jul 2019

Bikeman78 said:
Some examples of crazy things ARS does that a signaller would not. Below is a snapshot of trains at Cardiff Queen Street. Let's say 2F31 is six minutes late. It will roll into platform 3 heading south as 2D30 is waiting to depart platform 4 heading north. 2Y37 will be close behind 2F31 but cannot get across Queen St North Jn because the latter is blocking the platform. ARS won't let 2D30 go, it will hold it on a red until until 2F31 departs and 2Y37 arrives, probably putting four minutes delay into 2D30.

http://www.realtimetrains.co.uk/search/advanced/CDQ/2019/07/22/1205-1230?stp=WVS&show=all&order=wtt

Another example, if 2A24 is running late and departs Cardiff Central behind 2D30, there is every chance that ARS will hold 2D30 in platform 4 and route 2A24 around it via platform 5. The two trains take different routes from Queen St so all this does is put four minutes into 2D30 with no benefit to 2A24.

On the main line there are several routes that a passenger train from Newport can take to access platform 3 or 4 at Cardiff Central. If one train stops, e.g. it's booked platform is occupied, ARS will make no attempt to route subsequent trains around even if their booked platform is clear and there is a clear route available.

It’s like when late runners are ARS programmed to run DM to the East Junction then over the 25 crossings into P3 or P4.

Much better to as standard route DM over to the DR for P4 and DM to UR for P3 at Moorland Road which is a 75 Crossing, set both routes up from Moorland into the station this will then give both trains clear signals into the station with flashers at Moorland Road saves a good few minutes on the inbound.

My pet hate is the DR to UR approach control coming west, especially as ARS is programmed to cross freight onto Line D through the station. It’s quicker to route DR to Line C at Long Dyke then back from Line C to D at the West Junction, if required route the a passenger off P3 or 4 down Line E although the downside is the 25mph down E vice 75 down D.

Parham Wood · 21 Jul 2019

221129 said:
The alternative would have been a cancellation along the route due to drivers hours. Hence the run fast. The train failed to call at Cheltenham, Bristol Parkway, Taunton and Tiverton.

If only 4 mins late from Bristol Temple Meads why did it omit Taunton and Tiverton or is this a case of a decision having been made it cannot be unmade? I can see that in Cornwall there could be pathing delays if late which could lead to drivers hours being up but do they change drivers at Plymouth these days?

Bevan Price · 21 Jul 2019

vlad said:
Having checked RTT I'm amazed they let the Crewe train run its full route, as they tend to curtail it at Stoke if it's running late.

Even so, it arrived at Crewe after it was scheduled to depart (it's got a pretty tight turnaround) and didn't make up its delay completely till it got back to Derby.

Yes - I was a passenger on the 14:42 Derby - Crewe yesterday (20 July), a single Class 153 from Blythe Bridge to Crewe. (After a visit to the Foxfield Railway steam gala.). I think we caused a slight delay to the 16:01 Crewe to Euston (via Stoke & Birmingham) Class 350, while it waited for us to clear the single line section between Crewe & Bartholmey Jn.

So, any further delay to the Derby - Crewe train could have caused consequential problems as far as Birmingham or even London Euston.

Parham Wood · 21 Jul 2019

Perhaps we need a separate thread to discuss ARS etc.. Why do we need these automated systems at all? Does it make life easier for the signaller or mean a signaller can cover a larger area? If they need constant reprogramming to reflect operational changes that strikes me that they will always be behind the game. If they make silly routing decisions at times that cannot help either. Sorry these are a new concept to me.

Highlandspring · 21 Jul 2019

It allows one signalling workstation to control a larger geographic area thereby reducing the number of workstations required and saving £££s. The corollary being that whenever it has to be switched off, for example during degraded working, the signaller's workload can increase dramatically.

edwin_m · 21 Jul 2019

Parham Wood said:
Perhaps we need a separate thread to discuss ARS etc.. Why do we need these automated systems at all? Does it make life easier for the signaller or mean a signaller can cover a larger area? If they need constant reprogramming to reflect operational changes that strikes me that they will always be behind the game. If they make silly routing decisions at times that cannot help either. Sorry these are a new concept to me.

It's basically to reduce signaller workload and therefore allow fewer signallers to control a larger area. The idea is that ARS deals with things that are running routinely and the signaller can deal with problems by taking particular trains or particular parts of the area out of ARS control. It will also back off if a signaller cancels a route or it detects something like track circuits occupying and clearing out of sequence (suggesting a TC failure, for which it will also raise an alarm).

It's not completely dumb like the old piano roll systems they used to have on the underground, but is able to estimate trains' future behaviour based on their timetables and, if allowed to by its local regulating strategies, to select alternative routeings in an attempt to keep things moving. It can take account of timetable updates, including short-term alterations and those instructed by traffic management systems if they have the appropriate interface. But when I was involved in the early days the regulating rules were "hard coded" and needed input from BR Research to change them, and I suspect Resonate still has to be involved in this for the successor system they now own.

yorkie · 21 Jul 2019

Bevan Price said:
Yes - I was a passenger on the 14:42 Derby - Crewe yesterday (20 July), a single Class 153 from Blythe Bridge to Crewe. (After a visit to the Foxfield Railway steam gala.). I think we caused a slight delay to the 16:01 Crewe to Euston (via Stoke & Birmingham) Class 350, while it waited for us to clear the single line section between Crewe & Bartholmey Jn.

So, any further delay to the Derby - Crewe train could have caused consequential problems as far as Birmingham or even London Euston.

Maybe... or if your train had been any later, the Euston train may have gone first and not had to wait.

Parham Wood said:
Perhaps we need a separate thread to discuss ARS etc..

Feel free to create one, and link back to it here

Parham Wood · 21 Jul 2019

Highlandspring said:
It allows one signalling workstation to control a larger geographic area thereby reducing the number of workstations required and saving £££s. The corollary being that whenever it has to be switched off, for example during degraded working, the signaller's workload can increase dramatically.

Thanks. So it saves on signallers but is capable of howlers. I suppose signalmen are as well (we all are human) but it would appear not so often as the technology in some areas! I suppose eventually the technology / programming will improve, but I would ask in the S Wales case quoted above why the issues have not been addressed or is there a technology limitation?

Bevan Price · 21 Jul 2019

yorkie said:
Maybe... or if your train had been any later, the Euston train may have gone first and not had to wait.

True, but you are left with the choice of terminate the train short of destination - or risk delays at key locations between Crewe & Derby for several hours.

Tomnick · 22 Jul 2019

Tom Quinne said:
“Just let ARS do it” is a favourite, but I didn’t come into this job to watch a compute play Trains (badly) all day.

As you and others have also said, I used to find it immensely satisfying to see everything fall perfectly into place after a series of “ambitious” regulating decisions, so I find it all the more frustrating now as a driver when I’ve made the effort to make a load of time back only to be told “oh, the computer’s decided to let that out in front of you”. I know it always comes with the caveat that I can’t see the whole of the big picture or indeed what else is kicking off on the workstation, but there’s such a difference between when the ARS is clearly being left to get on with it and when the signalman’s getting stuck in.

Highlandspring said:
The Cardiff Main Line, Valleys and Vale of Glamorgan workstations are GETS MCS and the other workstations are WestCADs so they'll be using Hitachi's TREsa SARS (Signaller's Assistant Route Setting) rather than Resonate's ARS. The latter system is not too bad but the former makes some real howlers. Or at least it does in the locations where I'm familiar with its use.

I’ve never worked with any form of ARS directly (for which I am eternally grateful), but my perception here is that SARS in particular is utterly useless in anything worse than the mildest disruption, and chatting to others still in the signalling grade (who do work with it) only confirms that. My favourite remains being brought down to a stand from 70mph to allow (a couple of minutes later) a loaded freight train to emerge in front of me from a 15mph curve on an initially steeply rising gradient - knocked me for fifteen minutes, whereas I doubt it’d have even tickled the freight if I’d been allowed to run!

Mugby · 22 Jul 2019

Tomnick said:
I’ve never worked with any form of ARS directly (for which I am eternally grateful), but my perception here is that SARS in particular is utterly useless in anything worse than the mildest disruption, and chatting to others still in the signalling grade (who do work with it) only confirms that. My favourite remains being brought down to a stand from 70mph to allow (a couple of minutes later) a loaded freight train to emerge in front of me from a 15mph curve on an initially steeply rising gradient - knocked me for fifteen minutes, whereas I doubt it’d have even tickled the freight if I’d been allowed to run!

This, really, was the point of my original post.

The freight was allowed off the branch onto the Down Main, ahead of the 125, even though it was only going a distance of three miles or so before it would need to slow to 15mph to go on the loop at Clay Mills.
I don't buy the argument that it's load of concrete sleepers may have been crucial for some relaying job at 2pm on a Saturday afternoon.

Tomnick · 22 Jul 2019

Mugby said:
This, really, was the point of my original post.

The freight was allowed off the branch onto the Down Main, ahead of the 125, even though it was only going a distance of three miles or so before it would need to slow to 15mph to go on the loop at Clay Mills.
I don't buy the argument that it's load of concrete sleepers may have been crucial for some relaying job at 2pm on a Saturday afternoon.

There, though, it appears that the freight lost eight minutes between Stenson and Clay Mills, so the original margin might well have been adequate. Either way, though, if it didn’t run as booked, then it’d have to wait for the similarly delayed semi-fast behind yours and the Crewe behind that, then might struggle for a margin before the next fast which couldn’t be far behind by then, and before you know it, you’ve knocked it for an hour.

I have to say that, for the most part, they’re an excellent bunch in that establishment and it’s a pleasure to watch some of their creative regulating moves unfold!

Bikeman78 · 22 Jul 2019

Tomnick said:
My favourite remains being brought down to a stand from 70mph to allow (a couple of minutes later) a loaded freight train to emerge in front of me from a 15mph curve on an initially steeply rising gradient - knocked me for fifteen minutes, whereas I doubt it’d have even tickled the freight if I’d been allowed to run!

You've reminded me of a freight that went into Miskin up loop (as scheduled) for something like 30 minutes. During this time nothing overtook. The one train that was meant to overtake came up from Bridgend five minutes late so ARS let the freight out on time right in front of the passenger causing several minutes delay. My point being that the frieght could have come up to Cardiff early, and recessed west of Cardiff Central on line B if required, to give the passenger a clear run. For those not familiar with the area, line B is the one with signals 2240 and 2236 on this diagram.

https://www.opentraintimes.com/maps/signalling/cardiff#T_CRDFCEN

Tom Quinne · 22 Jul 2019

6B13 in the morning is the same Booked to recess at Miskin for the London to overtake, but if the freight is a little early (2/3 minutes) and the passenger a little late (3/4 minutes) you can run the freight to Cardiff, and get onto the Relief lines well ahead of the London.

Personally, I think the time it takes to slow, stop and start up again a freight like 6B13 could be saved (if there is a margin) if you let it run, but that means working your panel and not letting ARS run the show.

800002 · 22 Jul 2019

Tom Quinne said:
6B13 in the morning is the same Booked to recess at Miskin for the London to overtake, but if the freight is a little early (2/3 minutes) and the passenger a little late (3/4 minutes) you can run the freight to Cardiff, and get onto the Relief lines well ahead of the London.

Personally, I think the time it takes to slow, stop and start up again a freight like 6B13 could be saved (if there is a margin) if you let it run, but that means working your panel and not letting ARS run the show.

That is where the signallers knowledge and experience come into play. It is not unknown for individual pannels or signallers to have list of 'troublemakers' to watch out for.
ARS / SARS is only an aid, afterall, so it's only right that the signaller retain control of his / her trains.

What is annoying is when a train falls out of ARS due to an error in the signal, when it transfers to the adjecant desk / control area. Happens sometimes with schedules routed out of Reading via Southcote Jn onto the B+H desk, if the schedule error is around Newbury, it doesn't get routed at southcote (I believe) as ARS can't see where to put it.

Used to happen simillary at Westerleigh Jn area with the AXC ultra-long distance schedules too!

It does somewhat reduce the workload, under the right circumstances though, which I'd regard as a good thing.

A case of poor regulation?

Bikeman78

Established Member

RailUK Forums

vlad

Member

Tom Quinne

On Moderation

Parham Wood

Member

Bevan Price

Established Member

Parham Wood

Member

Highlandspring

Established Member

edwin_m

Veteran Member

yorkie

Forum Staff

Parham Wood

Member

Bevan Price

Established Member

Tomnick

Established Member

Mugby

Established Member

Tomnick

Established Member

Bikeman78

Established Member

Tom Quinne

On Moderation

800002

Member