News:

Use the "Forum Search"
It may help you to find anything in the forum ;).

Railroad signal operations?

Started by railfan727, December 06, 2010, 02:25:04 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

railfan727

Would it be possible to make the railroad signals work more like the real thing?  For instance,

1. To have the signals display Clear (green) aspects whenever a route is lined up, and not stay red until right before the train gets to it.
2. To have the signal turn back to red as soon as the locomotive passes it, rather than waiting until the entire train is in the block.
3. Simulate single track operations accurately, where trains are instructed to either stay on the main track or take sidings based on the train's priority (or whichever one gets to the siding first, take the siding if both are the same priority).
4. Be able to organize groups of signals into interlockings, and have them all function as a group to control traffic moves through a single switch, group of switches, or crossing at grade.
5. Display aspects other than just Stop or Clear, such as Approach, Approach Medium (or Approach Diverging), Medium Clear (or Diverging Clear), etc.
6. Distinguish between absolute (interlocking) and permissive (intermediate) signals.  On a single track line, whenever a route is lined from the end of a siding to the beginning of the next, all of the intermediate signals in between turn green in the anticipated direction of movement, and red in the opposite direction.

kierongreen

I think this is slightly outside the scope of simutrans really... The interface needed for these features would be overly complex to administer for large scale transportation networks.

prissi

1) Can be done. However, different countries have different systems, usually signale either display green or red all the time and do not switch until a train reaches a block. However, this disagrees with 2nd wish ...

2) can be done easily. (Either green default (as in old versions) or red default is both easily possible.)

3) use waypoints. Trains usually do not drive random routes in reality.

4) This is "read my mind" Usually crossings only block, when you use too many or double way signals.

5) I have a faint idea, what you are talking about. Did you consider, that those are very country dependent stuff? The only aspect, that is displayed, is the next block not empty in case of presignals.

6) Entire route? Then only one trains can go at all. One can think of eyecandy signals, that only indicate the state of reservation. But reserving the entire route defies the purpose of a signal.

Erik

1) How about two trains approaching a same part of track? Got one train suddenly red? ???
Here in Holland the signals before a switch are always red unless the train may go trough.
But the signals who hasn't a switch after them (only the next signal) are always green unless the track after them is occupied.


2) Should be great.

3) Mostly I solve this problem by making the siding based tracks less attractive for train who didn't belong there.
For instance: lower track speed, more turns and such. But some times even that wont work properly.
Making way points for all the trains (sometime more than 20) who passes that part of track is a lot of work.
Obviously if I make a change it's even more work.

4) Far as I understands it could be useful. Sometimes I have a couple of trains who blocks each other in a circle.

5) Far as I understands. Signals gets a third option. Who is letting pass a train with a lower speed if the next block isn't empty.
Something familiar was in discussion already approximately one year ago. But it is never finished.
http://forum.simutrans.com/index.php?topic=3713.0
(I really want this one.)

6) Sorry but I don't really see the advantage of it.
But it should be more beautiful that a signal turns only green in the direction the train anticipated to go, and red in the opposite direction.


ӔO

you can keep unwanted trains off the faster set of tracks by strategic placement of switch points.

The key thing to keep in mind is that all vehicles go as straight as possible when encountering a switch and will only change lines if that's the closer line to their destination or it's done through a waypoint. As such, when there is a branch line switch from a quad section, you can send the branch line into the slower track only and use a tunnel or bridge for the side that has to cross over.

If done right, you can make reduce the amount of waypoints that you have to set, drastically. I usually only need to use waypoints for express trains.
My Sketchup open project sources
various projects rolled up: http://dl.dropbox.com/u/17111233/Roll_up.rar

Colour safe chart:

railfan727

Erik,

The situation that you're describing there sounds to me like Rule 261 operation, in which trains may run in either direction on either track as necessary.  The signals at interlockings or "control points" (crossovers, junctions, and crossings at grade) are indeed always red unless the dispatcher has a route lined up.  The intermediate signals serve to provide spacing between following trains running in the same direction.  Only the signals at the crossovers are directly controlled by the dispatcher... when the dispatcher clears a route through a specific control point, all of the signals from that control point on to the next one all clear as well.

Let's use as an example CSX's double track Indianapolis Line in central Indiana.  In our example, we have an eastbound approaching CP 260 (crossovers) at Ingalls.  The next control point east of here is 10 miles ahead, at milepost 250 on the southwest edge of Anderson.  In between, there are three sets of intermediate signals, at MP 257, MP 255, and MP 253.  As our train approaches Ingalls, the dispatcher clears a route eastbound through CP 260 on track 1.  The intermediates at 257 and 255 would then display green, and 253 would be yellow (Approach indication) because CP 250 hasn't been cleared yet.  A short time later, a second eastbound is approaching CP 260 while the first one is in the block between MP 257 and MP 255.  The signal at CP 260 displays an Approach indication when the dispatcher lines the route.  As the rear of the first train passes the signal at MP 255, the home signal at CP 260 goes green, and MP 257 goes from red to yellow.

The crossovers can be used to get a faster train around a slower one, or in case one track is closed for maintenance.

On single track (which is far more common than double in the US), passing sidings are spaced several miles apart (the exact spacing depends on traffic levels).  In between sidings are intermediate signals, spaced about 2 miles apart, so that trains may follow each other in the same direction.  Each end of each siding is treated as a separate interlocking (control point).  When the dispatcher clears a route from one siding to the next, all of the intermediate signals in between turn green in the requested direction of travel, and red in the opposite direction.  Trains only run through the sidings if a meet is actually taking place there, and whichever train arrives first (regardless of direction of travel) is usually the one directed into the siding.  Otherwise they stay on the main track.  The signal at the facing point end of the siding has two or three heads (depending on the specific railroad's operating rules), and displays a Medium Approach or Diverging Approach if the switch is lined for the siding; this indication is red in the top head and yellow in the bottom (or red/yellow/red on a 3-header).  The last intermediate signal before a siding also has two heads, and displays green over red if the train is going straight through at the siding, or yellow over green (Approach Medium or Approach Diverging) if it is set to take the siding.

prissi

However, in simutrans terms, this means, that there would be eyecandy signals and real signal. And in german, default display is red unliek a free route is requested. It then depends usually on the length of the part of the route how many signals switch. Since this means read my mind, what a computer cannot do, it is out of question to make that reality. Either default green or default red signals are possible. Now "clever" sidlining, no other signals.

Please read my replies. (Also speed limit by signals are not possible with the current way speed limits are enforced and routes are searched.)

skreyola

These kinds of requests are fairly common. Perhaps someone needs to make a branch (a la ST Experimental) that incorporates more of these network-control abilities. :)
I'd probably play it. I like a little bit more hands-on, on occasion. :)
--Skreyola
You can also help translate for your language with SimuTranslator.

ӔO

I've occasionally wanted places with 3 sets of track, where the centre track is bi-directional and for overtaking, but it was too complicated to setup properly.
My Sketchup open project sources
various projects rolled up: http://dl.dropbox.com/u/17111233/Roll_up.rar

Colour safe chart:

VS

Deterministic pathfinding will always break that, no?

My projects... Tools for messing with Simutrans graphics. Graphic archive - templates and some other stuff for painters. Development logs for most recent information on what is going on. And of course pak128!

railfan727

#10
Quote from: prissi on December 07, 2010, 09:58:52 PM
However, in simutrans terms, this means, that there would be eyecandy signals and real signal. And in german, default display is red unliek a free route is requested. It then depends usually on the length of the part of the route how many signals switch. Since this means read my mind, what a computer cannot do, it is out of question to make that reality. Either default green or default red signals are possible. Now "clever" sidlining, no other signals.

Please read my replies. (Also speed limit by signals are not possible with the current way speed limits are enforced and routes are searched.)

The default display in the US is red too, under Centralized Traffic Control (CTC).  All signals are red until the dispatcher lines a route from one control point to the next.  Notice that I am not referring to the Simutrans definition of a route here... in Simutrans, this would be only the portion of a train's route between two consecutive interlockings.  Under Automatic Block Signal (ABS), the signals on a single-track mainline are green in both directions UNTIL a train passes the end of a siding onto the single-track section.  Then all of the intermediate signals in the opposite direction of travel from that siding all the way to the next siding "tumble down" to red.

Basically, what I'm getting at in my topic here is that it would be nice if there were a way to accurately simulate real-world CTC or ABS operation.

Paretje

For the single track, you should place a choose sign in the middle. If it's free, it will take the shortest track, otherwise it will switch. And, if both trains are in the same direction, the one can follow the other. But this doesn't work without a station.

railfan727

I'm able to get the trains through without any problem... it's just that the arrangement of the signals and the flow of the traffic is much different than what it is in the real world.  Rather than true single-track operation with passing sidings, my operation functions more like single-track stretches of dark territory interspersed with sections of double track operated under Rule 251.

ӔO

one of the hardest things to replicate in simutrans is the single track commuter branch line, especially when you have an express running on the same line.
My Sketchup open project sources
various projects rolled up: http://dl.dropbox.com/u/17111233/Roll_up.rar

Colour safe chart:

railfan727

That's why I always end up doing double track once I start running passenger trains...  if we had a way to tag trains with priority levels that would certainly help.  Lower priority, slow trains could be forced to wait in sidings to be overtaken by faster ones..

sdog

To get slow trains out of the way for express trains to overtake, i use two workarounds. (i avoid waypoints as they don't work on the reverse part of bidirectional schedules.)

For one i exploit that i use diesel local trains and electric express trains, the electrified track makes a little detour while the not-electric track doesn't. The diesels engines will go on that part, the electrics have to use the other track. When joining it's important to set up the signals right, to prioritize the expresses. (longer block lenght)

The other exploit is only possible in pak brittain, where i had heavy engines for the slow trains and light EMUs for expresses, selecting a track with a section with minimum weight below the heavy trains requirement causes the division.

skreyola

Quote from: Paretje on December 08, 2010, 11:11:20 PM
For the single track, you should place a choose sign in the middle. If it's free, it will take the shortest track, otherwise it will switch. And, if both trains are in the same direction, the one can follow the other. But this doesn't work without a station.
I think we need a new type of choose signal. It would act different from the station-area choose signal, in that it would only function if there was a rejoin signal on the path beyond, and it would only reserve routes to that rejoin signal, not try to reserve all the way to a destination station.
Or do you garbage up your schedules with waypoints?
I try to avoid waypoints at all costs.
--Skreyola
You can also help translate for your language with SimuTranslator.

railfan727

That sounds like a good idea.  But, if this signal is placed on the facing-point side of the switch at one end of a passing siding, could it be linked with the "rejoin" signals on the trailing-point side of the same switch (facing in the opposite direction) to allow movements past the interlocking in both directions?

And what about bi-directional intermediate signals to be placed at intervals on the single track between sidings to provide spacing for following movements, while the rejoin signals at the next siding prevent movements in the opposite direction until the entire single-track section is clear of traffic?  This would come close to simulating prototype operation under CTC with ABS.

skreyola

Perhaps it would have two types of signal: A Split signal and a Join signal, to differentiate from the Choose signal.
A Split signal would function like a normal signal unless there was downstream of it one of: 1) Another split signal, facing the opposite direction (to facilitate bidi track segments) or 2) a Join signal facing the same direction, which would mark the end of route reservation for the Split signal.
The Join signal could work like the signals in Railroads! in that it could make a train stop a couple of tiles back without an additional signal, so that a fast train could pass a slow train on a siding if it reached the Split signal before the slow train reached the Join signal... like this:

--S+--------TTTTT+J----
...\-------------/

S - Split
T - Train stopped by J
J - Join
--Skreyola
You can also help translate for your language with SimuTranslator.

railfan727

So, basically, we'd have a dual (or triple)-head signal at the facing point end of the switch, and a single-head signal for each track on the trailing-point side, facing in the opposite direction.  Just like would be seen with a typical CTC installation as seen in the US.

Is there a way to select whether a signal is placed on the left or right side of the track?  On a double-track railroad (or at the end of a siding), a signal on the right side of the left track would end up between the tracks.  The ONLY railroad that I know of that did their siding signals this way was the Wabash; they even went so far as to swing the end of the siding out away from the main, before curving back into the switch, to leave enough clearance on each side of the signal between the tracks.

There are two ways to avoid this: The vast majority of railroads simply place a standard ground mast to the left of the left track and the right of the right track.  But, back in the steam era, when long locomotive boilers prevented engineers from being able to see much of anything to their left (due to being seated on the right side of the cab), signals were commonly mounted either on overhead bridges, or on "bracket masts" situated to the right side of all tracks, with the order of the signals on the bracket corresponding to which tracks they govern.  The B&O, the N&W, and the Nickel Plate were heavy users of bracket mast installations.

skreyola

#20
Given that users can have rail lines that run-on-right or run-on-left, I don't think this bears too much discussion. I'd say put these signals on the same side as the other signals, or paint them as having a pole on that side but the signal over the track on an arm.
Come to think of it, isn't this a pakset dependent issue? However it's painted... so you could have paksets painted to have it on either side... or even both sides (positioned by Ctrl-button selection like orientation of stations?) or two buttons that link to wayobjects that are identical except for the graphic.
So, I don't think the side the signal sits on is a matter that needs serious thought at this point, since nobody's said anything about whether this will actually get programmed.
If you're thinking of making the graphic to spur the programming likelihood, I'd say, paint it both ways and let the pakset maintainer sort it out if it gets programmed.

Oh, and there should probably be a third signal I didn't realize we'd need, an S/J signal. It's a split to one direction and a join to the other. These sets of signals would need to be programmed carefully so they block the tracks in between in a failsafe manner, so you don't end up with a jam because of bidi operations, i.e., an S/J signal wouldn't allow blocking of a passing operation that would occupy all of the available sidings, so there's a place for traffic from the other side to go, or something.
This is probably pretty complicated, but I'm sure there's an elegant solution to be found.
--Skreyola
You can also help translate for your language with SimuTranslator.

Kirdan

I recently found OpenTTD and Simutrans and after some rather intensive study of the signal systems of both games I'll have to agree that they are both have problems with signals if you want to simulate real life train operations. And I think it's a shame, since it would be better games, if signaling was a solvable puzzle and not something which degenerates into "use the hack that works".

There's several things you could do to improve the simulation, but the most important problem is common to both: One-way signals means actually one-way.
My point being that in both games you cannot place a signal, which will only affect traffic in one direction. If you need to signal-functionality, you either have to accept that it applies to trains from both direction, OR you have to turn it into a one-way signal effectively closing the track for traffic in the other direction.
This is not realistic. I have a hard time recalling a rail-track which you can only run in one direction. Even non-bi-directional double track lines have signals for wrong-main-running. And they are independent signals. (In Denmark wrong-main-running will however turn the entire track to the next station into 1 block (unless it's a bi-directional track), but it is possible to use the left track).

I know a lot of signaling stuff is very country dependent, but I can't imagine it's so special to have signals which only affect traffic in one direction.

OpenTTD has "path-based signals", which does exactly this: Show red unless the entire track to the next signal can be reserved and DON'T affect traffic coming from the back side of the signal. It can be used to operate a single-track bi-directional line with tracks for crossing, but it still has some problems and don't interact well with the other signals. Other than that you are basically forced to use one-way signals and build bi-directional lines as double one-way tracks.

Has it been considered to allow 5 kinds of every signal:
* bi-directional
* affect-dir-A-block-dir-B
* affect-dir-B-block-dir-A
# affect-dir-A-ignore-dir-B
# affect-dir-B-ignore-dir-A

(*: existing, #: new)
???

Actually - in Simutrans I would suspect there's very little use for the existing one-way signals. You have the "Rail Close", which you could use to turn a track into really one-directional if you really meant it.


skreyola

@Kirdan: Given the way wayfinder works, I always use one-way signals in my networks. It's the only way to keep things running smoothly. Otherwise, you'd end up with two trains going opposite ways on a track that has no sidings between them, leading to deadlock.
I don't think your suggestion will happen. The only way it would have a slim chance of being included is if you sat down and coded all the changes that would have to be made to the wayfinder for it to be included, and that's a huge chunk of code. There's no guarantee it would be more efficient than the current method, and there's every reason to believe it would increase the processing requirements and break the way most players build their networks, with little benefit.
--Skreyola
You can also help translate for your language with SimuTranslator.

wlindley

I would contrariwise say there is relatively little use for the existing two-way signals, as if you put more than one of them on a stretch of track, trains will sit head-to-head waiting endlessly for the other to move.

"Real" railroad signals assume a dispatcher, or timetable operation.  Otherwise who decides which train should leave a station on a single-line road?  I cannot see how "real" signals would work in Simutrans... enlighten, please.

Lmallet

Quote from: wlindley on February 14, 2011, 02:25:27 AM
"Real" railroad signals assume a dispatcher, or timetable operation.  Otherwise who decides which train should leave a station on a single-line road?  I cannot see how "real" signals would work in Simutrans... enlighten, please.
I was going to say the same thing.  In reality, there are many types of signaling, and a railroad can use many types.  On this side of the pond, you have ABS, CTC, OCS, timetables.  Even the signals themselves can differ, as some display block occupancy, others can indicate speed or route.  Also railroads will often designate directions on multiple track lines.  While you can run a train in the reverse direction, this would only be done for exceptional cases.

I only have one use for two-way signals, and it usually is to protect a branchline.

Ters

There have been cases where I have really missed one-sided signals. They are dangerous since they can easily cause deadlocks if used without proper thought, but so are the two-way signals.

A typical use for such a signal in my games would be short, releatively rarely used sidings. I don't want trains leaving the station to start reserving tiles on the main line until it gets close, so I want a signal near the main line. This should be a one-sided signal ignored by trains entering the siding (they should reserve all the way to the station), but obeyed by trains leaving it.

Currently, my choices are:

  • a two-way signal with potential for dead-lock
  • a two-way signal with an intricate system of pre-signals and other signals to cancel out the effect of the pre-signals everywhere else
  • a waypoint with no visual representation
  • trying to make room for a double-tracked siding even for just a single train per month


-S<--------+S<-------S<
>S------->S+-------->S-
           S
           |            >S One-way signal allowing trains to move right
           |            S< One-way signal allowing trains to move left
           |            +  Junction
           #            S  Preferrably one-sided signal used by trains moving north only
           #            #  Station
           #
           #

Kirdan

#26
Thanks for the good arguments.
I do feel, however, that your statements about one-way signals confirm my argument.
Two-way signals are of little use and you preferably build one-way tracks with one-way signals (at least that's how I understood your comments).

However, one-sided signals can work. They partly do in OpenTTD in the form of path-based-signals (PBS).
A PBS is default red, and only turns green, if it can reserve a path to the next signal facing the same way.
This allows you to build single-track lines as:


         p__________q             p__________q
_________/p__________q\___________/p__________q\______
              A                        B


Notice the different facing of the signals (p, facing trains from right, q facing trains from left). A train leavning station A will wait at etiher of the "q" signals until it can reserve a free path to either platform on station B. Such a signal would actually do the job of "S" in Ters's post above.
Unless I'm mistaken, you can't build this with Simutrans long-block signals - excatly because they have to be either two-way or one-way.
The paradigm with PBS is that you place them at "the next safe waiting position". Using long-block signals in two-way mode, which risk making trains wait in the junction approaching stations. Using long-block in one-way mode would eliminate platform choice.

Speaking of choosing platform. Simutrans has the great feature over OpenTTD that you decide in the schedule which platform for a train to use. (in OpenTTD, you have to use via points to get a train terminating at a station to use a dead-end platform and not block the other trains going through).
But why is choosing platform a signal?! Shouldn't that be something you write in you schedule? And unless I misunderstood something you are forced to use the choose-platform signal in its two-way form, unless you want it to turn your line into a one-way line.

PS: Let's have some definitions to avoid confusion:
One-way: blocks traffic in the opposite direction
One-sided: ignored by traffic in the opposite direction.




@wlindley

You're right about two-way signals. What I meant about "very little use for one-way signals" was that you could get the same effect, by just closing a track section with a "rail close".
Wouldn't you be able to build the same networks as you do now with one-way signals, by just placing a two-way signal and a "rail close" ?
So removing the one-way signals in favor of one-sided signals wouldn't prevent network designs based on one-way tracks - that's what I meant.
 



Btw... I don't see the possibility for deadlocks as unconditional evil. There's different kinds of deadlocks.
Deadlocks able to happen by accident even when low traffic are bad, but deadlocks happening simply due to too much traffic for the line requiring manual intervention and expansion of capacity is not necessarily bad. Any network can deadlock with too many trains if you reach the inherent capacity. Even a purely one-way based design.
The above layout with station A and B would of course deadlock if station A had two trains waiting heading towards B and B had two trains waiting heading towards A. Manual intervention would be necessary to make one train turn around.
But that would happen in real life too. I've actually been is such a train on a rather large station (on a dual-track line) once, which had to go back, take another throughgoing track and go backwards into the platform to unload passengers, simply because a broken down train had reduced the station capacity below the traffic.





MOD Note: Please, do not double-post (i.e. post twice or more times within 24 hours). Use Modify instead. Thanks -Fabio

Sure... sorry. Will remember.

fbfree

I disagree with points 1, 2, and 5 as they are truly country and time dependent.  Points 1 and 5 have already been discussed.  For point 2, any train with a caboose or break van requires the conductor to verify the signal aspect.  For most of railroading history when this was the case, the signal aspect did not change until the train fully passed the signal.

3. It would be nice to have overtaking which the current system does not easily allow.  However, I do not see why on single track operation allowing trains to deterministically take either the left or right running siding is problematic.

4. Certainly for small interlockings, this is what pre-signals are for.  If you have a double track and single track segments coming into an interlocking, put a two-way signal on the single track segments and a one-way pre-signal on the incoming double track segments.  This keeps the interlocking clear. 

6. One important feature of the two-way signals in simutrans is the fact that the square with the signals can be reserved from either direction.  You can have trains follow (at least for one block) between passing sidings as long as the passing sidings themselves have enough storage capacity to clear the track (number of trains on branch - 1).  This only allows for one intermediate signal between passing sidings, but if you should need more than that, it's time to double track.

jamespetts

Something similar to no. 5 is under consideration for Experimental, albeit there are other priorities at present. The reason for this consideration is that the issue of braking distances and notice of upcoming signal aspects is an important determinant of the maximum speed and capacity of a line, which, in turn, has significant economic consequences. This is likely to be implemented alongside a more accurate means of modelling train braking rates. No. 3 can be done with waypoints: have a look at the demo.sve for Pak128.Britain for a good example of the effective use of a freight loop.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.