A problem to solve for the tough guys

You are here: Home > Forum > Simulations > Released > Carlisle > A problem to solve for the tough guys

Page 1 of 1

A problem to solve for the tough guys 26/11/2014 at 17:44 #65802
Finger
Avatar
220 posts
I have a problem for those of you interested in getting to the heart of things.

While playing the 10/11/14 timetable of Carlisle, I got myself into a deep muddle. The task is to get the sim into a "continuable" state, and describe the solution. Without removing trains, of course, and without sending them somewhere they don't belong, shunting, reversing, editing timetables is OK. Also, I'm interested in solutions that don't involve returning trains back to Carlisle.

The troubles are located at Dalston.

I've somehow got this to solve itself by reversing several times and changing the current location of the trains, but this was totally unexpected and I don't know what happened. I wonder if anybody can explain this behavior.

As a tentative solution, I lined up three trains behind each other on the Up Line at Dalston South, but for some (to me unknown) reason, I couldn't replace the ground frame levers to let the trains depart.

Any insight is appreciated. Happy Simsigging

Post has attachments. Log in to view them.
Log in to reply
A problem to solve for the tough guys 26/11/2014 at 19:49 #65805
TomOF
Avatar
452 posts
You have two trains facing each other in a Mexican stand off.

The timetable should be written using rules in such a way that a train will not try to enter unless another train has been in to drop off tanks.

Only one train can fit in the sidings at a time and also I think the sidings will not accommodate a 400M train.

I had to abandon timetable on the train heading in the down direction and reverse it, then get the other train out of the sidings.

Unfortunately a third train then entered which resulted in me not being able to normalise the ground frame which I had to delete.

So having 3 trains in the vicinity of Dalston is not prototypical, I'm afraid.

Hope this helps.

Post has attachments. Log in to view them.
Last edited: 26/11/2014 at 19:49 by TomOF
Log in to reply
A problem to solve for the tough guys 26/11/2014 at 21:20 #65808
Finger
Avatar
220 posts
" said:
The timetable should be written using rules in such a way that a train will not try to enter unless another train has been in to drop off tanks.

That's a worthy input for people developing timetables and even for fixing this one, thanks.


" said:

I had to abandon timetable on the train heading in the down direction and reverse it, then get the other train out of the sidings.

Meaning the first train was set loose to depart to Wigton? Personally, I prefer a solution that has predictable outcomes - in this case, I just set the stopping position of the arriving train so far that the departing train fits in the gap.


" said:

Unfortunately a third train then entered which resulted in me not being able to normalise the ground frame which I had to delete.

Yeah, but deleting trains is cheating. Still, what are the exact conditions for replacing the lever on the Dalston Oil ground frame? As I already indicated, I managed to get all the three trains on the Up line, still I couldn't replace the lever.

Log in to reply
A problem to solve for the tough guys 26/11/2014 at 21:41 #65809
TomOF
Avatar
452 posts
In order to normalise the ground frame there can't be anything either trying to leave the sidings or standing on the trackwork on the main line.

This is because there are dummmy signals controlling the exits from the siding and I wanted to avoid a situation where an adverse change of aspect could happen.

I think also in your save game I had to flip the levers so the 'outbound' lever was reversed allowing an exit from the sidings.

It was then that I got snookered by the third train immediately entering as soon as the second train made it out of the sidings.

Log in to reply
A problem to solve for the tough guys 27/11/2014 at 10:09 #65815
Finger
Avatar
220 posts
Finally, I managed to get all the three trains on the line and replace the levers. One of the rare occasions you can use the stopping position 1000 meters far .
Post has attachments. Log in to view them.
Log in to reply
A problem to solve for the tough guys 27/11/2014 at 11:20 #65817
peterb
Avatar
451 posts
Online
The easiest option for me was to remove 6C32 (yes I know, cheating...)

I don't think it helps that somehow you have trains which delayed by varying degrees trying to get in and out of the sidings at the same time, as Tom says three oil trains in the vicinity is not natural.

" said:
The timetable should be written using rules in such a way that a train will not try to enter unless another train has been in to drop off tanks.
Tom I don't understand what kind of timetable rules would apply for this. Also, I'm not sure if specific train rules would be effective in this case where the timetabled order of trains has become jumbled. Surely it is the case that the sidings should be coded in such a way that it is effectively OTW?

Also, on closer examination (google maps!) there appears to be two siding roads at Dalston, one ~100m long the other ~200m. Are they both in use? The train length of 400m therefore is obviously a timetable bug to be fixed but equally the sim surely should not be allowing trains of this length to easily fit in the sidings.

Log in to reply
A problem to solve for the tough guys 27/11/2014 at 12:38 #65818
Finger
Avatar
220 posts
" said:
Also, on closer examination (google maps!) there appears to be two siding roads at Dalston, one ~100m long the other ~200m. Are they both in use? The train length of 400m therefore is obviously a timetable bug to be fixed but equally the sim surely should not be allowing trains of this length to easily fit in the sidings.
It seems (from google maps) only the southern road can be used for unloading, and of it only the fenced part which is about 120 m. However, this photo shows two trains at Dalston, one using the other siding.

" said:
" said:
The timetable should be written using rules in such a way that a train will not try to enter unless another train has been in to drop off tanks.

I don't understand what kind of timetable rules would apply for this. Also, I'm not sure if specific train rules would be effective in this case where the timetabled order of trains has become jumbled.
Such rules would be constructed as follows:

Let's denote the trains T1, T2, ..., where odd numbers arrive at the sidings and even numbers run out. Let's assume there must always be at most one train in the sidings, T1 forms T2, T3 forms T4 and so forth. So, the obvious rules would be (1) T[2*n] must enter xx minutes after T[2*n-1] has left. To avoid deadlock at the reversing point, you would, on top of that, add a Dalston calling point for odd trains and add a rule (2) T[2*n+1] must leave Dalston 0 minutes after T[2*n] passes Dalston. This could still get you into a deadlock if eg. T3 ran before T1, which means there's a third set of rules needed (3) T[2*n+1] must enter 0 minutes after T[2*n-1] has entered. This should be fairly solid, meaning it could only break if one of the trains was lost, or overtaken by a following train.

BTW, why do the down trains call at Dalston? If that is needed, please give them a NX stopping position so they don't block the Up line needlessly.

And a second thought, since the trains operate with one loco (at least it seems so from the photos), IMHO they should be running round somewhere (not in the sidings, that's impossible). Dalston station (that would lmit the length of the train to about 140 m)?

Log in to reply
A problem to solve for the tough guys 27/11/2014 at 12:51 #65819
Peter Bennet
Avatar
5362 posts
" said:
" said:
Also, on closer examination (google maps!) there appears to be two siding roads at Dalston, one ~100m long the other ~200m. Are they both in use? The train length of 400m therefore is obviously a timetable bug to be fixed but equally the sim surely should not be allowing trains of this length to easily fit in the sidings.
It seems (from google maps) only the southern road can be used for unloading, and of it only the fenced part which is about 120 m. However, this photo shows two trains at Dalston, one using the other siding.

" said:
" said:
The timetable should be written using rules in such a way that a train will not try to enter unless another train has been in to drop off tanks.

I don't understand what kind of timetable rules would apply for this. Also, I'm not sure if specific train rules would be effective in this case where the timetabled order of trains has become jumbled.
Such rules would be constructed as follows:

Let's denote the trains T1, T2, ..., where odd numbers arrive at the sidings and even numbers run out. Let's assume there must always be at most one train in the sidings, T1 forms T2, T3 forms T4 and so forth. So, the obvious rules would be (1) T[2*n] must enter xx minutes after T[2*n-1] has left. To avoid deadlock at the reversing point, you would, on top of that, add a Dalston calling point for odd trains and add a rule (2) T[2*n+1] must leave Dalston 0 minutes after T[2*n] passes Dalston. This could still get you into a deadlock if eg. T3 ran before T1, which means there's a third set of rules needed (3) T[2*n+1] must enter 0 minutes after T[2*n-1] has entered. This should be fairly solid, meaning it could only break if one of the trains was lost, or overtaken by a following train.

BTW, why do the down trains call at Dalston? If that is needed, please give them a NX stopping position so they don't block the Up line needlessly.

And a second thought, since the trains operate with one loco (at least it seems so from the photos), IMHO they should be running round somewhere (not in the sidings, that's impossible). Dalston station (that would lmit the length of the train to about 140 m)?
I don't do a lot of timetable writing so with that caveat- if you want to avoid T1, T3 and T5 all entering then use the MutEx rule and if they are type Q (run as required trains) then make them all have less than 100% probability of running, that way you may get none entering but never more than one.

Peter

Peter

I identify as half man half biscuit - crumbs!
Log in to reply
A problem to solve for the tough guys 27/11/2014 at 13:19 #65820
Finger
Avatar
220 posts
" said:

I don't do a lot of timetable writing so with that caveat- if you want to avoid T1, T3 and T5 all entering then use the MutEx rule and if they are type Q (run as required trains) then make them all have less than 100% probability of running, that way you may get none entering but never more than one.
That's not the case. We want T1, T3 and T5 all enter, but only in that order. Not T3, T1, then T5, for instance. And never skip one of them, or the rule scheme would get in trouble.

BTW I find the NotIf rules pernicious - when I tried them, there was always some problem - either none of the trains ran, or it was always only the first, etc. The Are Alternatives rule seems to work much better for me (eg. on Peterborough).

Log in to reply
A problem to solve for the tough guys 27/11/2014 at 13:25 #65821
TomOF
Avatar
452 posts
In other words,
Trains can divide in the sidings but there will only ever be one locomotive because an incoming train is sometimes used to shunt tanks from one line to another or an incoming train might be split to enable it to fit.

Therefore the timetable must ensure that there will never be a situation where a train is trying to leave when another train is trying to enter.

Log in to reply
A problem to solve for the tough guys 27/11/2014 at 14:04 #65822
Finger
Avatar
220 posts
" said:
In other words,
Trains can divide in the sidings but there will only ever be one locomotive because an incoming train is sometimes used to shunt tanks from one line to another or an incoming train might be split to enable it to fit.

Therefore the timetable must ensure that there will never be a situation where a train is trying to leave when another train is trying to enter.

If there's ever only one locomotive for all the oil services, the rules would be even simpler.

Last edited: 27/11/2014 at 14:04 by Finger
Log in to reply
A problem to solve for the tough guys 27/11/2014 at 20:46 #65827
Finger
Avatar
220 posts
" said:
And a second thought, since the trains operate with one loco (at least it seems so from the photos), IMHO they should be running round somewhere (not in the sidings, that's impossible). Dalston station (that would limit the length of the train to about 140 m)?
Oh crud. This won't work because the sim doesn't allow such moves, particularly:


  1. Because there are insufficient track circuits at Dalston, a train stopped at the station will always lock one of the ground frames so they can't be switched. This is really a showstopper as any run-round at Dalsotn needs to move both GFs.

  2. And a slightly less important obstacle, the sim won't allow a train from Dalston South into the platform at Dalston. It will be held at an (imaginary) signal (can be passed at danger probably)


Log in to reply
A problem to solve for the tough guys 27/11/2014 at 20:53 #65828
TomOF
Avatar
452 posts
They do run round in the station, using one GF at a time. The length of trains is critical to avoid locking both GFs.
Log in to reply
A problem to solve for the tough guys 27/11/2014 at 21:10 #65830
Finger
Avatar
220 posts
" said:
They do run round in the station, using one GF at a time. The length of trains is critical to avoid locking both GFs.

Yeah, but as I said, this move can't be pulled off in the sim. Because there are only two track circuits on each line in the area for south GF, station and north GF, a train, however short, stabled at the station will lock at least one of the GFs. Particularly, a train at the down platform occupies T425 which blocks the southern GF, a train on the up platform occupies TC T398 which blocks the northern GF.

Log in to reply