ARS question

[Mike Stone]

I recently noticed on the open train times map a train arrived in platform 4 at Stafford. Normally the route behind it would disappear, but in this case it didn't.
;
Would this be due to ARS selecting the route for a following train?

[RDNA]

Post by Mike Stone » Mon Apr 17, 2017 3:07 pm

I recently noticed on the open train times map a train arrived in platform 4 at Stafford. Normally the route behind it would disappear, but in this case it didn't.
;
Would this be due to ARS selecting the route for a following train?

No Mike, currently ARS cannot 'overcall' routes, what you saw was probably caused by the signaller operating an AUTO facility for the next train to follow.

This lack of overcalling causes problems where signals on approach to a conflicting point have 'Double Red' controls (for TPWS overspeed purposes) i.e. the next signal in rear is approach controlled unless a 'live' route is set from the protecting signal.

DB

[rower40]

No Mike, currently ARS cannot 'overcall' routes, what you saw was probably caused by the signaller operating an AUTO facility for the next train to follow.

This lack of overcalling causes problems where signals on approach to a conflicting point have 'Double Red' controls (for TPWS overspeed purposes) i.e. the next signal in rear is approach controlled unless a 'live' route is set from the protecting signal.

DB

I've configured ARS to overset routes in specific cases where it allows the previous signal to clear to an unrestricted yellow instead of being approach-controlled-from-red. Great Eastern, Down Main through Seven Kings springs to mind as an example.
It's important to ensure that the route releases after one train before letting ARS set it for the next one, but the first train does not have to have left the section. Also ARS has to be configured to ignore the fact that the route has not proved within the usual 30-second route proving timer.
I agree that what was seen by the original poster was most likely the use of an A-button.

[Mike Stone]

Thanks - I had subsequently come to that conclusion, and wondered if it was being done to stop ARS crossing trains slow to fast through platform 3, with nothing approaching UF, which makes little sense.

[StevieG]

Not sounding like a similar situation in that no oversetting of route was involved, but in case of interest, just thought I'd mention about the ability of ARS to request the setting of route into an occupied terminal platform.

From the outset at Liverpool Street (1989), I recall that ARS was able to request the SSI to set Call-On class route into an occupied platform in at least two circumstances : -

- If the data for the approaching train in the TTP (TimeTable Processor) contained a next working's train description (TD), known as its "Association", which matched that already interposed for another train already in an accessible platform [whether done by ACI (Automatic Code Insertion) or manually by the signaller], then ARS would request the route into that platform for the second train, as an attaching of the two trains was apparently required ; OR

- If the TTP data of the approaching train included a "J" (Join) data instruction with the TD of another train already in a platform, then, again, route into the occupied platform would be requested by ARS.

Whether and at what point, the signal was permitted to display the call-on aspect for the second train was a different matter, separate from ARS operation, and dependent on the absence, or presence (and to what degree), of train-length measuring track detection sections in the platform(s) and at the signal's approach ('Lime Street' Control or similar), any provision of a 'Lime Street' override facility, etc.

Afraid I have no experience of ARS being configured to handle attaching / detaching or 'platform sharing' at through stations' platform lines, so do not know how such may be facilitated.

[Mike Stone]

Can ARS "change its mind"? At Stafford I frequently notice routes set and then drop out, sometimes twice, for the same train while it is approaching an approach controlled signal, which hasn't cleared?
;
The oddest one yet was 5A56, the Crewe-Wembley test train, initially routed into platform 5 on Friday, a move I have never seen done before, which then reverted to the usual platform 4. There was never any chance of it running through 5 without delaying the next down train.

[RDNA]

by Mike Stone » Sun Sep 10, 2017 5:43 pm

Can ARS "change its mind"?

No Mike, so far as I know ARS cannot cancel routes, it just sets them according to the conditions programmed in the "flowchart", often somewhat pretentiously called the "algorithm".

ARS does not have a 'mind', artificial intelligence has not been invented!

DB

[JRB]

Not a mind, but an algorithm could do it. They just have not invested in one.

[RDNA]

JRB » Sun Sep 10, 2017 8:58 pm

Not a mind, but an algorithm could do it. They just have not invested in one.

I beg to differ!

The sheer number of alternative actions and their consequences on a complex mixed traffic railway means that only an experienced human who actually cares about efficient train running can make the best educated judgement about regulating in the real world!

DB

[StevieG]

by Mike Stone » Sun Sep 10, 2017 5:43 pm

Can ARS "change its mind"?

No Mike, so far as I know ARS cannot cancel routes, it just sets them according to the conditions programmed in the "flowchart", often somewhat pretentiously called the "algorithm".

ARS does not have a 'mind', artificial intelligence has not been invented!

DB

==================================================================

After which, JRB posted :-
"Not a mind, but an algorithm could do it. They just have not invested in one."

I beg to differ!

The sheer number of alternative actions and their consequences on a complex mixed traffic railway means that only an experienced human who actually cares about efficient train running can make the best educated judgement about regulating in the real world!

DB

... which is why it may currently be a case of something like "We could do it but we don't".

Regarding Mike Stone 's most recent ("Can ARS "change its mind"?") query, unless things have changed on this point in the last ten or so years (which I much doubt : And provided that the Stafford control system is not different from IECC Classic in this respect), the answer to this query is (AFAIK) still in the name - Automatic Route-setting Sub-system : ARS cannot cancel a set route.

Barring a serious soft-or-internal hardware fault of some sort, [and apart of course, from Train Operated Route Release 'cancelling' sections of route behind the train], set but 'unused' routes can only be cancelled (remembering that, dependent on situation details, an Approach Locking time delay in releasing a route may apply) via the interlocking, and, AFAIK at present, this can only be by action of the operator, whether by cancelling the individual route, or use of a Signal Group Replacement Control or an interlocking's Emergency Signals On Control (or their equivalents).

ARS 'setting' routes does so only by making a "route request" (for that is the terminology; well, in IECC Classic anyway) to the interlocking : - Just as with a manual operator route-'calling' action (which, with an SSI, is also a "route request") : Unless an ARS-requested route is free to set in the interlocking, the request will elicit no response until the requested route does become free.

However the answer to 'Can ARS "change its mind"?' could be 'yes' if the situation was that where ARS is 'programmed' to cater for alternative routes or routing strategies, then, if an ARS Route Request is unsuccessful (perhaps after several repeat requests or after a given time period), or its strategy looks further ahead in comparing, say, degrees of presence or absence of trains on parallel lines (e.g. differing levels of queueing /'congestion'), ARS then might 'change its mind' before a requested route sets, by requesting the setting of a different route(s) for a theoretically better pathing.

Perhaps worth mentioning that, naturally, a route being 'not free' in an interlocking can be caused, not only by an already set route or train position being in conflict, but also if the operator has either locked any points in one position which a requesting route needs in the opposite position, or has applied a reminder to a signal or 'button'-equivalent to purposely prevent a route(s) from being set by any means.

Also that if there are any ARS Sub-Area controls (or equivalent) present, of which any one can then be 'turned off' to inhibit ARS from requesting route(s) from/in/on its(their) designated individual signal(s), area, direction, or line, such action does not prevent manual route requests.

So, regarding Mike's question about route appearing to be set and then cancelled ahead of a train, with no knowledge of circumstances, ARS algorithms' considerations, or signaller's thoughts, I can only speculate that either :-
- a route request (ARS or manual) was successfully made, and then manually cancelled, or
- in link(s) between the interlocking and by whatever means that the appearing and disappearing of route was observed, either :- -
- - the end display might differentiate between routes that are set ready for use, and those which have been cancelled but are 'timing out' or
- - route could have actually been set all the time but there may have been some 'hiccups' somewhere in the link(s).

[rower40]

Another take on 'Can ARS change its mind?':
In any priority decision between trains A and B, with train A timetabled first, there comes a value of train A's delay when ARS decides that minimum overall delay will occur by sending train B first. That value will be dependent on the exact timetables of trains A and B (including timetabled stops, pathing allowances etc), the geography of the junction where their paths conflict, and a host of other parameters such as the trains' classes.

Consider a hypothetical pair of trains A and B, timetabled 5 mins apart, where the above calculation comes out as "Train A has priority unless it is 7 mins late or more."

Train A approaches the junction 6m45 late, but for some other reason its route is not available. ARS doesn't set route for train B because train A has priority. But when train A starts to slow down approaching restrictive aspects, its delay trips over the 7m00 threshold, and ARS now sets route for train B.

ARS has over 25 reasons NOT to set route for a train, only one of which is "another train has priority". Only when none of those reasons apply will ARS issue the request to the interlocking to set the route. And all other trains in the ARS area are considered for priority decisions when determining if a route can be set. Many of these reasons change in real time, so ARS is continually considering all trains as to whether it should set a route for them.

[The Jazz Man]

To clarify, the ARS system at Stafford IS configured to overset a number of routes, these are all due to the double-red block protection afforded in the interlocking.

The ARS at Stafford is provided by Hitachi ('SARS') and has some differences to the 'old' IECC style ARS system, including number of enhancements which make it more powerful than IECC. It does not need to use Route Proving Exemption when oversetting as the proving timer doesn't start until all the tracks in the route (except overlap) are clear.

SARS can deal with joining and splitting of trains in any platform so long as the associations are defined in the CIF file (Timetable from NR).

To confirm what has been written above, ARS can't 'change its' mind' once a route has been requested - but it continually re-evaluates the regulation up to the point of any route-call. It certainly can't cancel a route once it is set.