Sideline into mainlinne

[audigex]

Benny - yes it is inefficient, I mentioned the problems in the post. It's not a replacement for every priority - but it's a useful extra to have in the right circumstances. For realism on a route where you don't need short ML seperation it's fine. I've seen several people who don't use the "signal every 2 tile" technique - preferring a signal every 2 train lengths or whatever. As with most things, it has it's place and doesn't fit everywhere. For those going for realism, however, it' can be very useful compared to the "what's that extra track doing there" method.

Petert... yes it will. Try it.

Edit: Oh, and if you make the last signal an exit pre-signal, and remove all but the first combo, you can increase throughput slightly.

[bwong]

um... wait, how do trains gain priority?

[petert]

um... wait, how do trains gain priority?

http://www.openttdcoop.org/wiki/Priorities

Petert... yes it will. Try it.

Edit: Oh, and if you make the last signal an exit pre-signal, and remove all but the first combo, you can increase throughput slightly.

Sorry, I read at the wiki link above that it would work.

[2457]

well, in your case (who asked the question) I do have that kind of situation quite often. As years have passed by I developed a quite handy solution for a problem that ideally should not exist. Quite a clever one, did take like 10 minutes to develop actually. What You need is a way to combine a load balancer and a line merger.

Simple it sounds, and actually simple it is. I show you in the form of this here picture.



Operation is quite useful. It will let trains merge with empty lines, main line has priority. So if both mainlines are occupied, trains wanting to merge will wait till they spot a proper opening on one of the 2 main lines.
IF both mainlines are free, and You have more merging trains they will go 1 to main line 1, and the next one will go to main line 2. Thats good, it adds a balanced load to both main lines. Keep an eye on signaling.
Pre-signals are essential to achieve proper working. The slope right at the entry signal is a good thing, if the train has to stop (both main lines are occupied) then it will use the slope later as a boost to acceleration. We benefit from faster merging, less chance that merging train will ever interrupt main lines, and it helps to keep the trains a little bit closer. combination with a cyclo injector is possible, but I never done it. It works almost flawless, there is only 1 exception when merging train picks one of the main lines, and a train on the main line enters the priority zone while the merging train is between the entry and exit signal. The merging train will stop, giving priority, but sadly if the other main line gets free it won't be able to choose that. This is a rare situation in my experience. And as far as i'm aware this is by far the best merging setup for merging 1 line into a dual mainline.

[audigex]

I think you can actually solve the problem with it stopping at the exit signal - you've just hit on (or possibly re-used) something I've never seen before with priorities - namely that there's no need for the combo's at all.

Get rid of all the pre-signals, and just put a standard one way signal at the split, and that should work.

I'll give it a go, but I can't see why it wouldn't work.

Edit:
After a quick investigation, it works for a single-line priority, but not in your situation it does in fact solve the problem of stopping to block one line, but it means the train will only go if BOTH lines are clear. ie ignore the above post

It does work for a single-line priority though. A priority can be made with a simple layout like below

pbsprio.png (31.38 KiB) Viewed 3752 times

Because the whole area can be a single block without effecting the PBS signals, the combo's aren't necessary and therefore neither is a start signal. I'm not sure if anyone's noticed this before.

[petert]

I think you can actually solve the problem with it stopping at the exit signal - you've just hit on (or possibly re-used) something I've never seen before with priorities - namely that there's no need for the combo's at all.

Oh yes, the pbs look ahead creates the same priority feature, easily and more simply.

I would prefer a solution like this one:

[2457]

well, i don't quite like mainlines with bridges. Mainline must go at full speed always.
also in your solution the block signals after the exit signals will give problems.

actually the priority tracks will trigger red signal on the block signals, but not on the exit signals.
so it ruins the whole purpose of the setup.

trains will not choose the free track.
they will pick one direction, even if the block signal is red, and simply wait there.

[planetmaker]

well, i don't quite like mainlines with bridges. Mainline must go at full speed always.
also in your solution the block signals after the exit signals will give problems.

actually the priority tracks will trigger red signal on the block signals, but not on the exit signals.
so it ruins the whole purpose of the setup.

trains will not choose the free track.
they will pick one direction, even if the block signal is red, and simply wait there.

PeterT's setup is absolutely correct and superior to yours IMO. For several reasons:

- you have a single tunnel on the ML. That ruins your signal density and will cause hickups. His bridges are doubled avoiding it. Also bridges usually don't have a speed limit lower than the fastest train. It is only important for the fastest maglev - or if you play without realistic acceleration.
- you have sharp bends on your side line. That ruins speed there.
- you have the problem of a side line train possibly stopping when it chose a side. That is much more of a problem than you said yourself, if you really have main lines which get used. It is bound to create big jams on your side line. His setup with one train length waiting space may have a single train on the SL wait possibly there. But usually the path finder finds the free way and for moderate main line usage it doesn't really matter anyway. But having a fluent ML is more important than having a train on the SL wait a bit. So... his setup doesn't ruin the purpose, it enhances it. If you feel like, you could by means of an extended priority also link that signal to the ML priority which leads to the one trainlength holding space after the side line splits.

[2457]

depends, my main line does not have to climb a slope.
thats an advantage.

if my sideline trains have to stop, the slope helps them accelerate faster when there is a nice spot open.
yet another advantage.

both designs employ 4 corners for the side line.


actualy our goals may differ. i prefer absolute undisturbed main lines.
if i would not, then giving priority would not be implented. simply have the pre signal setup determen the used mainline entry point, and to avoid jams on mainline allow a crossover between the 2 main lines.


best is if you set up a dummy game, and try the 2 setups your self. you will see the difference.

[planetmaker]

both designs employ 4 corners for the side line.

actualy our goals may differ. i prefer absolute undisturbed main lines.
if i would not, then giving priority would not be implented. simply have the pre signal setup determen the used mainline entry point, and to avoid jams on mainline allow a crossover between the 2 main lines.

My goal is also absolutely undisturbed mainlines - seems like the same. And I've played enough games on our servers to actually have seen the difference between your design and the one PeterT showed here. I challange you to join our PublicServer and test both designs in the end game phase on two different sideline hubs and compare. Given your love to optimize designs you should enjoy it It's not a big difference, but the waiting space for one train before the SL joins each of the ML is just the tiny bit one can get better over the direct split which you argue.

Also as stated: the bridges are no problem usually. A climb is not bad, trains will pass single slopes and bridges w/o speed loss. Otherwise you have the wrong engine (too weak) or wrong acceleration setting. On the other hand: a not doubled tunnel OR bridge (doesn't matter which) IS a possible jamming cause for main lines. Undisturbed main line means: signal distance always equivalent to two tiles or less, no steep slopes (single slopes don't hurt, no need for big terraform, depends on available engines), no un-prioritized side line joins, no un-balanced station entries and main line hubs.

EDIT: possibly one has to differentiate: under low to medium load on the main line, the direct split might have a slight advantage over the split with waiting space as the average waiting time for SL trains might be slightly lower. But as soon as the ML has heavy traffic, too, the join with waiting space is at a distinct advantage. I argue on the grounds that we always get ML which will be heavily used in the end game

Concerning the corners: Maximum speed of a train is determined by the length of track between two corners which turn in the same direction. For your setup only trains of length 2 tiles can pass unhindered w/o speed reduction to 111km/h at most (see http://wiki.openttd.org/Curve ) So there is a difference between 4 bends and 4 bends

[2457]

i usualy play witht he original acceleration model, so we might had found why our opinnions are different.
with the original acceleration model the slope gives quite a boost to acceleration, lol.


I usualy don't prefer the usual hub style construction.
my networks usualy have dedicated lines in eatch direction, but loaded and empty trains have a separate line too.

i prefer a tiny realism too, so i allway use feeders. less, but longer trains what i usualy have. (since every train has a gap to follow the train in front of it this increases the density the lines can take) and I so use depot buffers too, thats not realistic, but handy.

[petert]

Thank you planetmaker, I loved reading all your comments. To tell you the truth, I stole most of the design ideas from Mark's article, which you (2457) can check out at openttdcoop.org: http://openttdcoop.org/2009/10/05/joini ... g-chosing/

These designs bring up a good point, why do we use combo signals instead of PBS as priorities in the PublicServer games?

[planetmaker]

Thank you planetmaker, I loved reading all your comments. To tell you the truth, I stole most of the design ideas from Mark's article, which you (2457) can check out at openttdcoop.org: http://openttdcoop.org/2009/10/05/joini ... g-chosing/

These designs bring up a good point, why do we use combo signals instead of PBS as priorities in the PublicServer games?

- Normal block signals use less CPU
- "We" learnt to build priorities way before people thought about path signals and stick to our old ways.

- those reasons can basically summed up to "why not?"

And thanks for the comprehensive link with a discussion of both designs.

[Lord Aro]

since this has turned into a priorities disscussion, were prioities planned as part of the extra signals or were they just a happy accident?

[audigex]

Pre-signals were designed to allow a form of logic (binary, rather than philosophical), so in essence yes because they were added to allow more complex layouts. I don't believe anyone actually thought of priorities beforehand though, these things tend to be "discovered" later.

[Lord Aro]

could someone tell me what i'm doing wrong here?
i think it may be because of the really short trains, but i'm not sure

[petert]

Could you be a little bit more specific on the actual problem? If you're referring to the fact that trains can't get to the track to the right when the priority is red, use the following. It still needs to be fixed for Curve Length...

[Lord Aro]

i was actually talking about the fact that the trains coming in from the sideline stop the trains on the mainline

[petert]

i was actually talking about the fact that the trains coming in from the sideline stop the trains on the mainline

Was I supposed to guess?

Anyway, what do you mean they stop the trains on the mainline? When the block is full, of course the signal will be red...

[planetmaker]

I think I see it now what you mean (you could really use more description), my guess is the shift main line to the right of the image: you need tracks separated by a tile and a signal. Otherwise, of course the train on the lower ML stops, if the following combo signal on the upper ML is red or the parallel track section is used... Search the #openttdcoop wiki for shift main line. Pay attention to the details when re-building it. Differences between one-way and two-way signals matter in this case.