Optimizing 2 Way Trains
-
Rarefushion
- Manual Inserter
- Posts: 2
- Joined: Sat Aug 14, 2021 3:59 am
- Contact:
Optimizing 2 Way Trains
I've been working on a two way train system you can see here. I want to make it as efficient as one way and easy for others to use. Long track should be easy if you just add wait splits, I haven't tested this theory, but for tighter systems it's much slower than one way. It might be impossible but I'm looking for help with this goal.
Last edited by Rarefushion on Mon Aug 16, 2021 12:17 am, edited 1 time in total.
Re: Optimizing 2 Way Trains
That's an impressive collection of blueprints !
I fear for you that there is no way to make two-way-one-track system as easy as one-way-x-track, because you can make a functionnal or a totally broken mess out the same blueprints. But much more easily with the 2-way-1-track system.
If i try to think of a two-way-one-track, then cut it into pieces to make it modular similar to what is linked, there would be:
1)single sections
1A)straight
a) size in rails or chunk or train size
b) orientations
1B) with angle/ curve
a) size in rails or chunk or train size
b) orientations
2) dual sections ( waiting bay for crossing )
2A)straight
a) size in rails or chunk or train size
b) orientations
2B) with angle/ curve
a) size in rails or chunk or train size
b) orientations
3) loading stations
4) unloading stations
5) loops
6) junctions
7) stackers ( before stations)
Undetailled category seem to me pretty similar to what you already have for 1-way-x-track. just a quick note on 6) loop, it could be added or not to station blueprint but also at other location; it is required for train to turn around obviously, because you network most likely will not be a circle with all station on it. They need to be at least the sized of the larger train.
I wanted to stress out the fact that to have a complete set of blueprint one would need to have available many different kind of dual section to allow for trains to manoeuvre that's the most important/difficult i think.
Those pieces, how many of them , and their locations require a bit of thinking to corellate with the amount of trains and network structure. Much more than in single-way-x-track network where they do not exists.
If you try to simplify as much as possible the network structure and train schedule, some rules of thumb that can be applied automatically could emerge i think to make the use of blueprint easier. but that need to be done considering different people have different approaches. ( like having a network that is just one giant circle with all station on it would be uncommon for me so my "rules" are not useful in this case, and probably others i didn't thought/saw).
For me personnaly, i would simplify by having every train doing A-B-A-B schedule. where A and B are stations followed by a loop to make a dead-end.
In this case such rule could be something like : from every (pointA) to every (pointB) you need to have at least [(number of train doing A-B)-1] dual sections, the size of the larger train.
if you have Mine A, furnace B,and 1 train, everything fine.
if you have Mine A, furnace B,and 2 trains, one need to step aside for the other to pass. If it's done at a station by putting it on a large-enough-for-2-trains-loop and not using a dual section that mean 1 train is forced to stay at a station at all time.
if you have Mine A, furnace B,and 5 trains, for this, you will need at least 4 waiting bay located between A and B. (=space for 4 trains in dual section).
Now you can place them wherever you want(= use as many straight pieces from category 1 between them and the stations) it should work but with various/variating efficency of the network.
If you have less waiting bay, it means there will always be 1 or more train forced to stay at a station at all time.
Maybe i'm doing a mistake here but i would compare it to those 4x4 or 5x5 and so on square puzzle game where there is one piece missing that allow you to move adjacent pieces and you need to shuffle them around to make an image visible when all pieces have their correct positions.
(Edit: https://en.wikipedia.org/wiki/Rush_Hour_(puzzle) similar logic ).
The more "free space" the easier it is to resolve the game, the faster the train will resolve itself.
That's only one aspect on a very simple network. If you add junctions, it adds a constraint on the location of the waiting bay because between the junction and a station there can only be [1 train + number of waiting bays ]. A waiting bay itself, a merge, a X cross, all those are junctions.
Now if your trains do A-B-C-A-B-C schedule. It makes it more complicated to think ahead how many dual section you need and their location is made also more difficult to define if you have another train doing D-C-D-C, even more so if that train don't do that when A is not empty, because in this case A is closed and E and F opens and the train that was doing D-C-D-C now does D-C-E-F-D-C-E-F until C runs dry and close and so on.
How many "free space" is the minimum between B and C, if all trains that "could" end up between B and C, ends up being between B and C ?
Now place this amount between the correct junction considering all trains that ended up doing B-C travel are taking every different path available to them and make sure no other trains will utilize any of it on their way from A to D for example.
At this point if you are still confident into branching stuff here and there using blueprints you may be forcing yourself not to use a 1-way-x-track system. And the thought process before adding things may take longer than necessary while the building part itself doesn't increase all that much. ( spending 60 minutes to plan ahead for 20 rails may be you fun).
But that's more where your already existing blueprint dictionnary shine. those don't have the similar use case imo.
2-way trains could be a nice addition on some less busy area, small patches here and there to deliver supply on wall, or service small mine, or taxi thing, but very rarely as the main core of a big network
( i saw you already had the junction for branching those single track into "regular" network of train, maybe with little signal and track rework you could include them in the dictionnary )
Also on early game, when steel production is slow and you don't have much for rails, not as to be made with bots, but manually at the correct size so to be upgraded later into a 1way-x-track system. That's a different thought-process of making "optimized 2 way trains". Which could also be a nice addition to your dictionnary, as an early stage of it.
that's some "theoric thought" but i've also used them quite a lot for fun because i wanted to make them work. I wouldn't say "impossible" but a different magnitude of complex to replicate the same functionnality. However that may not be the wisest move to try and replicate the same functionnality. again (imo).
I fear for you that there is no way to make two-way-one-track system as easy as one-way-x-track, because you can make a functionnal or a totally broken mess out the same blueprints. But much more easily with the 2-way-1-track system.
If i try to think of a two-way-one-track, then cut it into pieces to make it modular similar to what is linked, there would be:
1)single sections
1A)straight
a) size in rails or chunk or train size
b) orientations
1B) with angle/ curve
a) size in rails or chunk or train size
b) orientations
2) dual sections ( waiting bay for crossing )
2A)straight
a) size in rails or chunk or train size
b) orientations
2B) with angle/ curve
a) size in rails or chunk or train size
b) orientations
3) loading stations
4) unloading stations
5) loops
6) junctions
7) stackers ( before stations)
Undetailled category seem to me pretty similar to what you already have for 1-way-x-track. just a quick note on 6) loop, it could be added or not to station blueprint but also at other location; it is required for train to turn around obviously, because you network most likely will not be a circle with all station on it. They need to be at least the sized of the larger train.
I wanted to stress out the fact that to have a complete set of blueprint one would need to have available many different kind of dual section to allow for trains to manoeuvre that's the most important/difficult i think.
Those pieces, how many of them , and their locations require a bit of thinking to corellate with the amount of trains and network structure. Much more than in single-way-x-track network where they do not exists.
If you try to simplify as much as possible the network structure and train schedule, some rules of thumb that can be applied automatically could emerge i think to make the use of blueprint easier. but that need to be done considering different people have different approaches. ( like having a network that is just one giant circle with all station on it would be uncommon for me so my "rules" are not useful in this case, and probably others i didn't thought/saw).
For me personnaly, i would simplify by having every train doing A-B-A-B schedule. where A and B are stations followed by a loop to make a dead-end.
In this case such rule could be something like : from every (pointA) to every (pointB) you need to have at least [(number of train doing A-B)-1] dual sections, the size of the larger train.
if you have Mine A, furnace B,and 1 train, everything fine.
if you have Mine A, furnace B,and 2 trains, one need to step aside for the other to pass. If it's done at a station by putting it on a large-enough-for-2-trains-loop and not using a dual section that mean 1 train is forced to stay at a station at all time.
if you have Mine A, furnace B,and 5 trains, for this, you will need at least 4 waiting bay located between A and B. (=space for 4 trains in dual section).
Now you can place them wherever you want(= use as many straight pieces from category 1 between them and the stations) it should work but with various/variating efficency of the network.
If you have less waiting bay, it means there will always be 1 or more train forced to stay at a station at all time.
Maybe i'm doing a mistake here but i would compare it to those 4x4 or 5x5 and so on square puzzle game where there is one piece missing that allow you to move adjacent pieces and you need to shuffle them around to make an image visible when all pieces have their correct positions.
(Edit: https://en.wikipedia.org/wiki/Rush_Hour_(puzzle) similar logic ).
The more "free space" the easier it is to resolve the game, the faster the train will resolve itself.
That's only one aspect on a very simple network. If you add junctions, it adds a constraint on the location of the waiting bay because between the junction and a station there can only be [1 train + number of waiting bays ]. A waiting bay itself, a merge, a X cross, all those are junctions.
Now if your trains do A-B-C-A-B-C schedule. It makes it more complicated to think ahead how many dual section you need and their location is made also more difficult to define if you have another train doing D-C-D-C, even more so if that train don't do that when A is not empty, because in this case A is closed and E and F opens and the train that was doing D-C-D-C now does D-C-E-F-D-C-E-F until C runs dry and close and so on.
How many "free space" is the minimum between B and C, if all trains that "could" end up between B and C, ends up being between B and C ?
Now place this amount between the correct junction considering all trains that ended up doing B-C travel are taking every different path available to them and make sure no other trains will utilize any of it on their way from A to D for example.
At this point if you are still confident into branching stuff here and there using blueprints you may be forcing yourself not to use a 1-way-x-track system. And the thought process before adding things may take longer than necessary while the building part itself doesn't increase all that much. ( spending 60 minutes to plan ahead for 20 rails may be you fun).
But that's more where your already existing blueprint dictionnary shine. those don't have the similar use case imo.
2-way trains could be a nice addition on some less busy area, small patches here and there to deliver supply on wall, or service small mine, or taxi thing, but very rarely as the main core of a big network
( i saw you already had the junction for branching those single track into "regular" network of train, maybe with little signal and track rework you could include them in the dictionnary )
Also on early game, when steel production is slow and you don't have much for rails, not as to be made with bots, but manually at the correct size so to be upgraded later into a 1way-x-track system. That's a different thought-process of making "optimized 2 way trains". Which could also be a nice addition to your dictionnary, as an early stage of it.
that's some "theoric thought" but i've also used them quite a lot for fun because i wanted to make them work. I wouldn't say "impossible" but a different magnitude of complex to replicate the same functionnality. However that may not be the wisest move to try and replicate the same functionnality. again (imo).
-
Rarefushion
- Manual Inserter
- Posts: 2
- Joined: Sat Aug 14, 2021 3:59 am
- Contact:
-
JimBarracus
- Filter Inserter
- Posts: 365
- Joined: Mon Jul 03, 2017 9:14 am
- Contact:
Re: Optimizing 2 Way Trains
I think the biggest downside is that the intersection area is one block.
Only one train can pass the intersection.
Only one train can pass the intersection.
Re: Optimizing 2 Way Trains
Honestly, it I think that moving to two one-way tracks IS the optimization called for here.
Either you are trying to time things so that you get material just in time as needed, or you upgrade the single two-way track to a pair on one-way tracks. And in the face of what upgrading to the latter offers, the former is neither time not cost effective.
Either you are trying to time things so that you get material just in time as needed, or you upgrade the single two-way track to a pair on one-way tracks. And in the face of what upgrading to the latter offers, the former is neither time not cost effective.