ARS at Woodbrough Loops

You are here: Home > Forum > Miscellaneous > The real thing (signalling) > ARS at Woodbrough Loops

Page 1 of 1

ARS at Woodbrough Loops 02/11/2019 at 13:39 #121449
lazzer
Avatar
600 posts
Following a discussion with a fellow driver the other day I was wondering if anyone knows if the Woodborough Loops on the Berks and Hants route have ARS. By this I mean (assuming ARS is provided on the panel in the first place) would the computer have the ability to automatically route a freight train into the Down Woodborough loop and then automatically route it out the other end with no manual input from the signaller?

I assume that such a move would require the system to be programmed to recognise a train that should be looped and route it accordingly. Is that actually possible, or do looping moves have to be made manually by the signaller?

Log in to reply
ARS at Woodbrough Loops 02/11/2019 at 13:52 #121450
JamesN
Avatar
1570 posts
Yes, the Newbury workstation is equipped with ARS; and ARS will route trains into the loops at Woodborough if scheduled; and back out at or approaching departure time.

Edited to add:-

If a train is NOT scheduled to be recessed at Woodborough; left to its own devices ARS will not loop the train, if the signaller wanted to loop a train in these circumstances they would have to manually intervene.

Last edited: 02/11/2019 at 13:54 by JamesN
Reason: None given

Log in to reply
ARS at Woodbrough Loops 02/11/2019 at 14:15 #121451
lazzer
Avatar
600 posts
JamesN in post 121450 said:
Yes, the Newbury workstation is equipped with ARS; and ARS will route trains into the loops at Woodborough if scheduled; and back out at or approaching departure time.

Edited to add:-

If a train is NOT scheduled to be recessed at Woodborough; left to its own devices ARS will not loop the train, if the signaller wanted to loop a train in these circumstances they would have to manually intervene.
That makes sense, as said driver found themselves on yellows down towards and past Woodborough the other day, so called the signaller.

"Sorry driver - the ARS routed a late-running freight into the loop and then routed it straight back out in front of you".

Makes you wonder what the signaller was doing to allow that to happen, given that the driver then had to follow that freight all the way to Westbury.

Log in to reply
ARS at Woodbrough Loops 02/11/2019 at 14:20 #121452
jc92
Avatar
3626 posts
Online
lazzer in post 121451 said:
JamesN in post 121450 said:
Yes, the Newbury workstation is equipped with ARS; and ARS will route trains into the loops at Woodborough if scheduled; and back out at or approaching departure time.

Edited to add:-

If a train is NOT scheduled to be recessed at Woodborough; left to its own devices ARS will not loop the train, if the signaller wanted to loop a train in these circumstances they would have to manually intervene.
That makes sense, as said driver found themselves on yellows down towards and past Woodborough the other day, so called the signaller.

"Sorry driver - the ARS routed a late-running freight into the loop and then routed it straight back out in front of you".

Makes you wonder what the signaller was doing to allow that to happen, given that the driver then had to follow that freight all the way to Westbury.
supposition here based on SimSig experience but he was probably not expecting it to do so, or was looking at something else when it happened and couldn't risk an ACOA by putting back.

"We don't stop camborne wednesdays"
Log in to reply
ARS at Woodbrough Loops 02/11/2019 at 15:10 #121453
JamesN
Avatar
1570 posts
Don’t forget the Newbury signaller is looking after a lot of route miles - in-fact I’m fairly sure it’s the highest route mileage of any of the TVSC desks. While it’s comparatively simpler railway vs Paddington or Reading; there’s more “faff” to deal with just because of the greater length.

That’s why ARS was provided - so the signaller can deal with shunting at Theale; be taking/giving up a line blockage at Hungerford all while not bringing the railway to a stand at Woodborough.

Log in to reply
ARS at Woodbrough Loops 02/11/2019 at 15:30 #121454
lazzer
Avatar
600 posts
jc92 in post 121452 said:
lazzer in post 121451 said:
JamesN in post 121450 said:
Yes, the Newbury workstation is equipped with ARS; and ARS will route trains into the loops at Woodborough if scheduled; and back out at or approaching departure time.

Edited to add:-

If a train is NOT scheduled to be recessed at Woodborough; left to its own devices ARS will not loop the train, if the signaller wanted to loop a train in these circumstances they would have to manually intervene.
That makes sense, as said driver found themselves on yellows down towards and past Woodborough the other day, so called the signaller.

"Sorry driver - the ARS routed a late-running freight into the loop and then routed it straight back out in front of you".

Makes you wonder what the signaller was doing to allow that to happen, given that the driver then had to follow that freight all the way to Westbury.
supposition here based on SimSig experience but he was probably not expecting it to do so, or was looking at something else when it happened and couldn't risk an ACOA by putting back.
I suppose my gripe with this is that the freight would have either come out onto the Down Main from Theale or at Reading, so the signaller would have known full well it was running. He may even had looped it at Towney beforehand, which should have made him even more aware of its presence.

As a driver I look at it one way, but as a Simsig player I can appreciate how difficult it is to keep an eye on everything all the time. But on balance, I would say the signaller dropped a clanger on this one, and the delay minutes are definitely going to Network Rail.

Log in to reply
ARS at Woodbrough Loops 02/11/2019 at 16:57 #121455
GeoffM
Avatar
6274 posts
Online
jc92 in post 121452 said:
lazzer in post 121451 said:
JamesN in post 121450 said:
Yes, the Newbury workstation is equipped with ARS; and ARS will route trains into the loops at Woodborough if scheduled; and back out at or approaching departure time.

Edited to add:-

If a train is NOT scheduled to be recessed at Woodborough; left to its own devices ARS will not loop the train, if the signaller wanted to loop a train in these circumstances they would have to manually intervene.
That makes sense, as said driver found themselves on yellows down towards and past Woodborough the other day, so called the signaller.

"Sorry driver - the ARS routed a late-running freight into the loop and then routed it straight back out in front of you".

Makes you wonder what the signaller was doing to allow that to happen, given that the driver then had to follow that freight all the way to Westbury.
supposition here based on SimSig experience but he was probably not expecting it to do so, or was looking at something else when it happened and couldn't risk an ACOA by putting back.
I would agree with that supposition. The loop either has no wait-for-time flag set in its data, or is set and the freight train was on time / late.

Only curious thing is that Western loop exit signals are usually approach controlled, as in needing the berth track occupied, to reduce the risk of a potential read-across from the main line. So the freight would have had to have already crossed the pointwork into the loop before the signaller would risk an ACOA to the freight, so probably a good minute or more. But as has been said, the signaller could have been dealing with an incident elsewhere.

The signaller will probably blame ARS: I used to see it all the time when we installed simulators in the IECCs, when in fact the signaller was actually inattentive, whether intentionally or not. It's an easy scapegoat that has no ability to defend itself.

Going back to SimSig, I get a lot of tickets for "ARS wrongly put X in front of Y". Well maybe it did, but it's not perfect and neither is the real world ARS (whether IECC or TRESA). We don't want to make it too easy for the users...

SimSig Boss
Log in to reply
ARS at Woodbrough Loops 03/11/2019 at 13:36 #121462
Late Turn
Avatar
696 posts
lazzer in post 121451 said:
JamesN in post 121450 said:
Yes, the Newbury workstation is equipped with ARS; and ARS will route trains into the loops at Woodborough if scheduled; and back out at or approaching departure time.

Edited to add:-

If a train is NOT scheduled to be recessed at Woodborough; left to its own devices ARS will not loop the train, if the signaller wanted to loop a train in these circumstances they would have to manually intervene.
That makes sense, as said driver found themselves on yellows down towards and past Woodborough the other day, so called the signaller.

"Sorry driver - the ARS routed a late-running freight into the loop and then routed it straight back out in front of you".

Makes you wonder what the signaller was doing to allow that to happen, given that the driver then had to follow that freight all the way to Westbury.

My recent perception is that the preference at some boxes is/was to sit back and let ARS do its thing as much as possible, and I've heard suggestions that they've been told (in one ROC) that it shouldn't be turned off or overridden without a very good reason. Maybe it's a policy from the local management to try to minimise the delay put onto them by letting ARS take it instead - I don't know. It's certainly a big step up from a tool that the signalman can use to help keep one end of the job running by itself for a while whilst he concentrates on some drama at the other end, to something that lets him sit back and switch off from it altogether (which is surely inevitable when you let the computer do *all* of the thinking for you - I've certainly always found it harder to supervise someone or something else than to do it myself!). I have noticed a change in recent weeks though, so maybe there's been a bit of a shift in policy or maybe it's just that the signalmen are getting more used to the quirks of ARS and working out what it's likely to do next - accounts from others who have worked with it for years suggest that it's like constantly supervising a particularly inept trainee?!

In this specific case, presumably the ARS firstly pulled off into the loop because that's where it was booked to go, then secondly pulled off straight out of the loop because it decided that the late-running freight should have priority over the passenger for whatever reason. Leaving aside the wisdom of the second choice, the two decisions are made independently so the computer wouldn't be considering which train would run first beyond the loop outlet at the moment that it pulled off for the freight to go inside - if I've understood it correctly!

I've

Log in to reply
ARS at Woodbrough Loops 04/11/2019 at 08:22 #121473
sorabain
Avatar
72 posts
I've got no knowledge or experience with this stuff, but interested in automation.

I was wondering whether the freight was only "slightly" delayed and passenger train ended up delayed by the freight (as it's been said it was running on yellows). The freight gets looped as per the timetable, but ends up close to its designated departure time anyway (perhaps it was there to let this passenger train pass if on time), then ARS decided that since the passenger train was now late (albeit through no fault of its own) and the freight was now on time then the freight got priority?

Log in to reply
ARS at Woodbrough Loops 06/11/2019 at 07:22 #121506
flabberdacks
Avatar
575 posts
In most places I know of, any delays caused by trains sitting at reds while ARS is disabled, the signaller cops it. It's easy to become distracted by a phonecall or radio or a micromanaging Control or anything else happening around you. A lot of signallers refuse to turn it off even when they really should, instead using other means to control it, such as locking points.

Also, having ARS makes for a lazy signaller workflow. If an area never gives you any trouble, it's only human nature that you would get used to that and your situational awareness will drop. Those of us who have worked fast and busy manual boxes know the level of situational awareness you need to not make a mistake - signallers who get too comfortable under ARS don't watch as closely, don't know the timetable as well, don't know the ins and outs of their interlocking well enough, they get caught out much more easily.

Log in to reply
The following users said thank you: GeoffM, ozrail
ARS at Woodbrough Loops 10/11/2019 at 21:15 #121570
Mikehax
Avatar
31 posts
sorabain in post 121473 said:
I've got no knowledge or experience with this stuff, but interested in automation.

I was wondering whether the freight was only "slightly" delayed and passenger train ended up delayed by the freight (as it's been said it was running on yellows). The freight gets looped as per the timetable, but ends up close to its designated departure time anyway (perhaps it was there to let this passenger train pass if on time), then ARS decided that since the passenger train was now late (albeit through no fault of its own) and the freight was now on time then the freight got priority?
In relation to the automation and the prioritisation of trains by ARS, the calculation of which train to run is based on three main factors - Class of train, Minutes early / late and type of junction. It is important to note that ARS does not differentiate between a train being early or late when making this calculation in that the calculation for a train that is 10 minutes late produces the same result as that for a train that is 10 minutes late!!

A value is assigned to each of the three variables:
Class of train - Class 1=15, Class 2=10 Class 4, 5 or 6 = 5
Timing - Value equals number of minutes early or late
Type of Junction -

Converging crossing \
\
-------- =50
/
/

Simple Crossing ----------------
\
---------------- = 100
\ \
\ \

Single (BiDi) Line \ /
\ /
--------------- = 150
/ \
/ \

The training module for signallers at York ROC gives three scenarios for how ARS would calculate priority:

Scenario 1.
Train A, a Class 1 (10 minutes late) and Train B, a Class 2 (7 minutes late) approach a converging crossing.

Calculation applied by ARS -
Train A = 15x10x100 = 15000
Train B = 10x7x100 = 7000

So Train A runs first

Scenario 2

Train A, a Class 1 (2 minutes late) and Train B, a Class 6 (12 minutes early) approach a simple crossing.

Calculation applied by ARS -
Train A = 15x2x50 = 1500
Train B = 5x12x50 = 3000

So Train B runs first - Clearly not the sensible choice but resulting from the inability of ARS to differentiate between early and late!!

Scenario 3

Train A, a Class 1 (15 minutes late), Train B, a Class 5 (20 minutes late) and Train C, a Class 2 approach a single / bi-directional line (A & B from one end and C from the other).

Calculation applied by ARS -
Train A = 15x15x150 = 33750
Train B = 5x20x150 = 15000
Train C = 10x10x150 = 15000

Train A runs first, Train C runs second and Train B runs third (Train C runs before Train B due to higher weighting of Class of Train as overall totals are equal)

Hope this is of some help in seeing where signaller intervention is required especially in relation to ARS decisions regarding early running trains.

Last edited: 10/11/2019 at 21:21 by Mikehax
Reason: Attempts at diagrams were even more rubbish than they are now!

Log in to reply
The following user said thank you: Soton_Speed
ARS at Woodbrough Loops 11/11/2019 at 08:54 #121582
sorabain
Avatar
72 posts
Mikehax in post 121570 said:
...
Hope this is of some help in seeing where signaller intervention is required especially in relation to ARS decisions regarding early running trains.
I'm a bit surprised by this. I thought that after a certain amount of delays priority was given to non-delayed trains. A bit disconcerting if some very delayed freight might get put in front of faster local services, especially with the "delay" scoring factor being multiplicative rather than additive.

Is there some minimum value for "early/late", at least non-zero? It would seem odd to give a train a score of zero if it was "on time" and anything else gets non-zero. Even with this, the multiplicative nature seems likely to make bad decisions regularly e.g. a class 1 one minute late (score 1x15 ignoring junction type) to have a lesser score than a class two 2 minutes late (score 2x10 ignoring junction type).

What role does the junction type play exactly? In your examples it was always equal for both trains since every train is conflicting over the same junction. I'm not sure when it would discriminate between two trains with conflicting routes. Is it if there are multiple junctions close together to pass over and one train's route may only pass over one whereas another might converge after the first and only pass over one? I can't quite picture that working as it would seem odd to give priority (again multiplicative so would dominate train class) to those passing over the most complicated route.

Log in to reply
ARS at Woodbrough Loops 11/11/2019 at 09:39 #121584
Mikehax
Avatar
31 posts
sorabain in post 121582 said:
Mikehax in post 121570 said:
...
Hope this is of some help in seeing where signaller intervention is required especially in relation to ARS decisions regarding early running trains.
I'm a bit surprised by this. I thought that after a certain amount of delays priority was given to non-delayed trains. A bit disconcerting if some very delayed freight might get put in front of faster local services, especially with the "delay" scoring factor being multiplicative rather than additive.

Is there some minimum value for "early/late", at least non-zero? It would seem odd to give a train a score of zero if it was "on time" and anything else gets non-zero. Even with this, the multiplicative nature seems likely to make bad decisions regularly e.g. a class 1 one minute late (score 1x15 ignoring junction type) to have a lesser score than a class two 2 minutes late (score 2x10 ignoring junction type).

What role does the junction type play exactly? In your examples it was always equal for both trains since every train is conflicting over the same junction. I'm not sure when it would discriminate between two trains with conflicting routes. Is it if there are multiple junctions close together to pass over and one train's route may only pass over one whereas another might converge after the first and only pass over one? I can't quite picture that working as it would seem odd to give priority (again multiplicative so would dominate train class) to those passing over the most complicated route.
As a signaller, all we are told is what is in the training module. However, I do know that there are Delay Rules applied by ARS which although pre-programmed can be manually amended for a particular area / work-station such that if a train is more than a certain number of minutes late or early it's level of priority is reduced although these only apply to Class 1 and Class 2 trains as others already have the lower weighting.

With regard to the junctions, again I am not 100% certain but I think that you are correct in that previous / further junctions will also be taken into account. ARS calculates train priority from the strike in point (usually about half way through the previous work-station / box's area of control) to the strike out point (usually half way through the next work-station / box's area of control).

I agree that in isolation the individual calculations seem rather odd.

Last edited: 11/11/2019 at 09:43 by Mikehax
Reason: Correction of typos

Log in to reply
ARS at Woodbrough Loops 11/11/2019 at 17:36 #121599
GeoffM
Avatar
6274 posts
Online
Unfortunately that ARS training package is, to put it kindly, simplifying things somewhat. Either that or they've ripped out the planned non-IECC ARS equipment. Source: I led the development of Signaller's Assistant ~10yrs ago which worked similarly to IECC ARS, and was supposed to be used for Sheffield but had no involvement in the (what turned out to be) extremely lengthy safety case.

So when a train has fewer than two green signals ahead, it considers the route it needs from the red signal. It will compare itself with every other train in the system. Most will be rejected because they're irrelevant, but for the ones where it does find a conflict, it scores the conflict for each train individually. It does come up with a score but more on that in a moment. The train with the best score is given priority: if it's the train being considered for routesetting then - all other things aside - it will now request the route. Otherwise this train is put into the "giving priority to X" state and no further action is taken until the next cycle (about a second later).

The conflict is determined by finding the limits of the conflict. A simple crossing could be considered as a diamond crossing - both trains have different entry and exit signals and share at least one track section. A forward working conflict is where at least one common signal is shared between two trains. A reverse working conflict is similar but the trains work in opposite directions over two or more common track sections. Then there are also conflicts in overlap (which have to take into account overlap release time for one or both trains, as applicable). There are five types of conflict. If a train is predicted to stop because of a regulating decision then a start-up allowance is also added.

The scoring is started by working out how long each train will take to enter the fouling point, and vacate the clearing point. This is simply by interpolating the timetable's timing points and the route lengths (which can be tweaked if there are major speed changes). The class of the train is considered which has a weighting factor for each class - some areas they are all the same; others vary a lot. Next, the actual delay to each train is also considered (and these have thresholds, so a train +/- 2 minutes of on time is "on time"). Finally the two numbers (one for each train) are normalised to avoid runaway numbers messing up the calculations.

The route under consideration may have variations such as a warner route, a calling-on route, or more than one path between the same pair of signals. Each of these is also considered and scored separately, so while a main route may cause delay to the other train, a warner might not so that could be selected instead.

There are a lot of tweaks that can be made, such as the train classes I already mentioned, plus the thresholds for being on time, plus individual route weightings (rare, but outbound routes in terminal stations with a single track throat may be more preferable than inbound, for example), number of green signals ahead of a train to maintain, and tidal flows (time of day dependent weightings). Then there are rules for allowing trains to leave early, and/or waiting for TRTS, or some other custom rules.

Despite all that, it's a ton of maths that can and does get skewed by factors outside its control or knowledge. That's why the emphasis is always on assistance and not replacement for signallers.

I'm sure I've missed a lot out!

SimSig Boss
Last edited: 11/11/2019 at 17:53 by GeoffM
Reason: None given

Log in to reply
The following users said thank you: Steamer, sorabain, Mikehax, postal, trolleybus
ARS at Woodbrough Loops 11/11/2019 at 20:06 #121606
Mikehax
Avatar
31 posts
Thanks Geoff. That makes more sense than the training module. And yes, Sheffield and Rotherham workstations are both still using the original SARS.
Last edited: 11/11/2019 at 20:09 by Mikehax
Reason: None given

Log in to reply
The following user said thank you: GeoffM
ARS at Woodbrough Loops 14/11/2019 at 22:20 #121649
MrBitsy
Avatar
121 posts
I have worked the Newbury panel and have had ARS do something daft as you describe. As far as the signaller 'allowing' it to happen, think about it - we would have to check what ARS is going to do with EVERY single train! If we had to do that, may as well turn it off!

Experience tells us when ARS may do something odd so we can catch it. In your example a reminder appliance can be put on the exit signal to stop ARS routing something straight back out. If a train doesn't need to go into the loop, we can clear the main line straight past - ARS cant override a manually set route.

The signaller may have just taken an emergency call, be dealing with a fault or failure or be on the phone arranging a line block. ARS is there as an aid to give the signaller breathing room for all the other tasks that need to be done. There are many reasons why ARS may make an error like yours. Maybe the schedule data was wrong for example.

TVSC Link 4 signaller - Temple Meads, Bath & Stoke Gifford
Log in to reply
ARS at Woodbrough Loops 14/11/2019 at 22:34 #121650
MrBitsy
Avatar
121 posts
If ARS made the mistake, why do you think the signaller should take delay minutes - I guess you would be annoyed if you personally received delay minutes if you loco failed! Any delay caused is obviously investigated and Train Planning may get the delay if they put a wrong line code in somewhere.

As I said elsewhere, The only way guarantee no ARS errors is to turn it off. Can you imagine the amount of time would be spent spent checking the schedule and interrogating ARS for EVERY train!! It is rather like you as a driver dismantling the brakes before each trip just to be on the safe side!

I was on panel 3 at West Hampstead in the morning rush a few years back. A late running freight was given an up path at 8am (normally refused until 9am at Bedford). It came to a stand on the up slow at St. Albans, causing considerable grief to me and the passengers. Driver contacted me with the news that the loco had run out of fuel. I have no idea why that happened, could have been a faulty gauge or other piece of equipment. Maybe the person who should have filled the tank, didn't. Maybe the engine had a fuel leak? I certainly didn't assume the driver 'allowed' it to happen :-)

TVSC Link 4 signaller - Temple Meads, Bath & Stoke Gifford
Log in to reply
ARS at Woodbrough Loops 14/11/2019 at 22:55 #121652
MrBitsy
Avatar
121 posts
flabberdacks in post 121506 said:
In most places I know of, any delays caused by trains sitting at reds while ARS is disabled, the signaller cops it. It's easy to become distracted by a phonecall or radio or a micromanaging Control or anything else happening around you. A lot of signallers refuse to turn it off even when they really should, instead using other means to control it, such as locking points.

Also, having ARS makes for a lazy signaller workflow. If an area never gives you any trouble, it's only human nature that you would get used to that and your situational awareness will drop. Those of us who have worked fast and busy manual boxes know the level of situational awareness you need to not make a mistake - signallers who get too comfortable under ARS don't watch as closely, don't know the timetable as well, don't know the ins and outs of their interlocking well enough, they get caught out much more easily.
Depends on the signaller I guess :-)

We have been trialling Train Management software at TVSC for a year. Some signallers don't use it it AND turn ARS off, but I use both. I will see a conflict 20 minutes out and fix it in train management. I'll change lines, junctions and/or times and my edits are passed on to ARS to carry out. I know the kind of mistakes ARS may make, so I'll amend the schedule to take care of them.

An example may be GWR ringing through a set swap that requires a platform change, split or join. I will open up a scenario in Traffic Management (snapshot of the service) and change the schedule while talking to GWR. I can see other conflicts that will be generated by the change, so can fix those at the same time. Whem GWR and myself are happy, I can synchronise the live service into my scenario to check nothing else has changed while planning. I then submit those changes that update ARS. I then get on with other tasks safe in the knowledge ARS will carry out the new plan.

Of course you could do all this manually, but forget to do it because you are dealing with an emergency call :-) Yes, any automation can make any of us a bit lazy, depends on the individual :-), but I find the new tools a great aid and engaging to use.

TVSC Link 4 signaller - Temple Meads, Bath & Stoke Gifford
Last edited: 14/11/2019 at 22:56 by MrBitsy
Reason: None given

Log in to reply
ARS at Woodbrough Loops 14/11/2019 at 23:58 #121654
JamesN
Avatar
1570 posts
Pesky GWR Service Controllers asking for very last minute swaps at Reading.....
Log in to reply
ARS at Woodbrough Loops 15/11/2019 at 09:00 #121659
MrBitsy
Avatar
121 posts
JamesN in post 121654 said:
Pesky GWR Service Controllers asking for very last minute swaps at Reading.....
:-) Bristol Temple Meads too!

TVSC Link 4 signaller - Temple Meads, Bath & Stoke Gifford
Log in to reply