YAPP/PBS Basics

[kaan]

The distance will be longer, yes.
But it often happens in my games, that when the signals are too close to each other (and so do trains) then on heavily used tracks,trains have problems reaching their maximum speed, because they constantly run into train ahead of them. Not once I was able to solve that by removing some signals.

A question of balance as usual

The only case where I can see that happening is when the train in front is slower that the train behind it.
Signals every two tiles works flawlessly for me.

[FunMaker]

It really depends on the network itself.
If you got a 10 Line Mainline and no more room to extend it and the traffic is really high on every line then you should consider to put some more signals on the rails (because it
makes more trains possible inthe same part of a track).Further more the chance that space for another train appears(cause the train allready on the line are grouping) is much higher

If there are few trains on a line it's not needed to put more signals than every train-length-tile.

in Short: many signals if you are going for high traffic and fast sl->ml joining. few signals if you are going for speed on the mainline and doesn't have much traffic.

btw: My Signal Length is 1 Tile and most of the time the speed is at max. But Mchl is right about the "running into a train ahead of them" and that might continue backwards to the next train, to the next train and so on until there is space between 2 trains to buffer this problem.

@kaan In my games it mainly happens if a train joins the mainline while not at full speed and a train on the ml can't switch to another line and is at full speed.

[DaleStan]

The most efficient layout is a signal every other tile and no tunnels or bridges without double track.

If every third tile is better than every fourth tile, and every other tile is better than every third tile, then why is every tile not better than every other tile?

[planetmaker]

The most efficient layout is a signal every other tile and no tunnels or bridges without double track.

If every third tile is better than every fourth tile, and every other tile is better than every third tile, then why is every tile not better than every other tile?

You're right for straight tracks - but consider the looks!

Another aspect are tunnels and bridges. For signals every tile, you'd have to tripple a bridge already, if it was three tiles long. With signals every other tile, you need start to tripple bridges only, if their length exceeds 6 tiles - which is sufficient for bridging other tracks and many other cases.

[Youri219]

The most efficient layout is a signal every other tile and no tunnels or bridges without double track.

If every third tile is better than every fourth tile, and every other tile is better than every third tile, then why is every tile not better than every other tile?

When multiple tracks merge or split you will not be able to keep up with the signal at each tile anyway.

[DaleStan]

When my trains come to junctions, the fact that the train is slowing down to turn has a whole lot more effect on the throughput than the fact that I don't have signals every tile in a junction. If smooth motion at junctions is the object then there should be signals every third tile, at most.

And I still fail to see why a long section of straight track, with no junctions, has more capacity with signals every other tile than with signals every tile.

[SoLo]

And I still fail to see why a long section of straight track, with no junctions, has more capacity with signals every other tile than with signals every tile.

With no junctions then every tile will be more efficient.
But how often do you have routes with multiple trains and no junctions? (aside from maybe the starting trains)
The problem occurs that when there is a junction at some point in the track you suddently have a spot that is not every tile and if there happen to be two trains running right behind each other, when they hit this spot the second train will have to stop.

[Rainer]

Hi,

The most efficient layout is a signal every other tile and no tunnels or bridges without double track.

If every third tile is better than every fourth tile, and every other tile is better than every third tile, then why is every tile not better than every other tile?

From my experience, "every tile" adds no efficiency (because even a single switch breaks the system). I cannot opt for a three tile distance, because I never tried it. When I tried out different distances (about five years ago) I switched from four directly to two and experienced a boost in throughput on heavily used tracks.

see you
Rainer

[T-Unit]

You must remember that if you are using an even signal distance it is best to have your trains an odd length. Therefore they can fit between the signals.

[Merlon]

You must remember that if you are using an even signal distance it is best to have your trains an odd length. Therefore they can fit between the signals.

Though , that only applies if you're using autosignaling on your diagonal tracks, since autosignaling adds half a hex to the signaldistance on diagonals. With a bit of OCD and time, you can keep the exact same signallenght even on diagonal tracks thereby keeping things optimized for one trainlenght.

[planetmaker]

You must remember that if you are using an even signal distance it is best to have your trains an odd length. Therefore they can fit between the signals.

Though , that only applies if you're using autosignaling on your diagonal tracks, since autosignaling adds half a hex to the signaldistance on diagonals. With a bit of OCD and time, you can keep the exact same signallenght even on diagonal tracks thereby keeping things optimized for one trainlenght.

On diagonal tracks a train is longer though - so the autoplacement isn't too bad. See diagonal bug

EDIT: fixed link (thx Ammler)

[audigex]

If you're using a gap for a whole train after the junction, you're limiting your train seperation to one train length.

A slightly more elegant solution is to divide your train length by desired seperation, then create that number of "out" tracks, each with one train length at the start. Then merge them after a couple of train lengths. The merge should work out okay (no jams) since you're using a divison of a train length. Obviously this gets prohibitive with huge trains, but for 8 tile trains with 2 tile seperation, you just need to provide 4 exits.

Not perfect, but it helps improve throughput.

[Timitry]

Even on diagonal tracks odd trainlenghts fit quite good with signals each 2 tiles - see http://www.openttdcoop.org/wiki/User:Tim/Tilelength

And yes, signals every 2 tiles should be the best solution, signals each tile is not possible due to junctions and each three tiles or four or five lowers the throughput.

[audigex]

One other thought - you don't ALWAYS have to leave a train length gap after a junction....

When MERGING two tracks into one, the trains will all be going to the same place - there's no chance of any train blocking another, so don't use the gap afterwards (it's a waste, the trains are following each other anyway)

When SPLITTING one track into two, the trains are all coming from the same place - so again there's no chance of blocking each other and again there's no need for a gap afterwards.

Where there are two or more entrances, AND two or more exits, leave at least one train length afterwards. If there is another split before a train length, don't signal until after the next split.

I'll get some screenshots made up later to demonstrate

[Noldo]

When SPLITTING one track into two, the trains are all coming from the same place - so again there's no chance of blocking each other and again there's no need for a gap afterwards.

What if one of the routes has is jamming and the other is not and the trains arriving to the split are all going to the non-jamming lane?

[Brianetta]

Having a signal block longer than one tile is important if you ever experience a decent jam. It's easier to turn train around and sort them out if you have a little room for the train to move backwards.

[audigex]

When SPLITTING one track into two, the trains are all coming from the same place - so again there's no chance of blocking each other and again there's no need for a gap afterwards.

What if one of the routes has is jamming and the other is not and the trains arriving to the split are all going to the non-jamming lane?

I said if ONE line goes to TWO - they wouldn't get through regardless of whether you use PBS or normal signals.

The first train would just wait further back, before the split, instead of blocking on it - it would actually take longer to clear since it would have to travel further.

If the first line has two tracks, that fits into the 2+ to 2+ tracks, and doesn't count - maybe you're thinking multi-line main lines.

[AlgoJerViA]

Does anyone have any good tips for 4 lane PBS layouts? This is my current setup:

One of the problems I encounter is that trains seem to just randomly switch lanes. This can lead to problems, because sometimes one train switching lanes will block a train, who will have to stop, only to switch lanes itself, blocking another train. In a busy line this can go on for a while, and this is why you won't see switches that permit both left-right and right-left lane switching in my screenie.

Now what I would like is for my trains to stay in the right lane whenever possible, and then switch to the left if needed, e.g. when overtaking. Any ideas for this? I tried putting opposing ("wrong way") signals on the left lane as a penalty, but sometimes trains rather stop than take the opposing signal.

It is covered in the manual.
http://wiki.openttd.org/index.php/PBS#A ... ack_layout


I have played with this for a while and it works very well for cargo that does not get punished for very long transit time like coal and iron ore since at first the traffic flows like nothing else the trains always tries to stay in the outer track but can use the two in the middle if needed so brakedowns and so are no problem at all. However when there is a lot of trains they will stop and wait for a free track cloging up the two side tracks and form a nice line of 20 or so trains because a train just leaving the station decided to use the third track the next 100 lines or so. But there is no real jams and all trains does get through sooner or later in a relatively even order.

//AlgoJerViA

[el koeno]

I have played with this for a while and it works very well for cargo that does not get punished for very long transit time like coal and iron ore since at first the traffic flows like nothing else the trains always tries to stay in the outer track but can use the two in the middle if needed so brakedowns and so are no problem at all. However when there is a lot of trains they will stop and wait for a free track cloging up the two side tracks and form a nice line of 20 or so trains because a train just leaving the station decided to use the third track the next 100 lines or so. But there is no real jams and all trains does get through sooner or later in a relatively even order.

//AlgoJerViA

It doesn't look that different from the quad track lay-out I posted. An as far as I can see my problem with it -trains switching lanes causing a chain of lane switching and waiting- hasn't been addressed.

[audigex]

If you put directions facing both directions on the two inner tracks, you can increase throughput. The outsides are free, so no deadlocking. It also has the advantage of penalising the centre tracks, so trains will try to keep to the outer tracks unless overtaking/manouvering.