4-way intersections: Throughput and deadlocks [image heavy]

Smart setups of railway stations, intelligent routing, solutions to complex train-routing problems.
Please provide - only if it makes sense of course - a blueprint of your creation.
User avatar
Odhrean
Long Handed Inserter
Long Handed Inserter
Posts: 76
Joined: Mon Sep 15, 2014 7:49 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by Odhrean »

aaargha wrote:Added Chunk-aligned cross. I think that my tests may be a bit heavy on the right hand turns, as this one preformed surprisingly well, or it's the insane acceleration from the rocket fuel that does it.
Thank you. The relatively good performance surprises me aswell but i think this cross is good enough where it makes sense.
Factorie
Fast Inserter
Fast Inserter
Posts: 100
Joined: Wed May 31, 2017 2:29 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by Factorie »

Holy hell this is insane. I barely understand the railway system in general.
aaargha
Filter Inserter
Filter Inserter
Posts: 333
Joined: Wed Dec 07, 2016 8:35 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by aaargha »

Just a quick update on the multi-lane test bench: I've managed to piece together a mod that allows for automated train deployment which, basically, allows me to create as many trains as needed to always keep the input lanes saturated without needing to have storage space, or schedules, for 300 or so trains. Users also won't need to change all the trains if they want to compare say, 1-4-1 vs 2-4, or 1-1 vs 5-25, just changing a few trains that are used as templates should be enough. It's a bit clunky to work with but at least it will be possible to make those kinds of comparisons without dying of boredom.

Hopefully I'll have some time to actually publish the mod, and integrate it into the test rig to run some tests, in a few days.
gyro2death
Burner Inserter
Burner Inserter
Posts: 14
Joined: Mon Aug 29, 2016 3:14 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by gyro2death »

aaargha wrote:Just a quick update on the multi-lane test bench: I've managed to piece together a mod that allows for automated train deployment which, basically, allows me to create as many trains as needed to always keep the input lanes saturated without needing to have storage space, or schedules, for 300 or so trains. Users also won't need to change all the trains if they want to compare say, 1-4-1 vs 2-4, or 1-1 vs 5-25, just changing a few trains that are used as templates should be enough. It's a bit clunky to work with but at least it will be possible to make those kinds of comparisons without dying of boredom.

Hopefully I'll have some time to actually publish the mod, and integrate it into the test rig to run some tests, in a few days.
I would actually enjoy such a mod if you do release it. Out of curiosity can you test a BP of my own?

This is my ultra-compact (as physically small as you can manage with my track distance) 4 way intersection. It doesn't support NW/SE despite my damn best efforts. I struggled long and hard to make it work but had to expand eventually and led myself to the Celtic Knot, though my signaling wasn't as good as their version you have here. Still would like to know what throughput I lose to save the extra 2 train tiles. Also my trains are just 2-4 at the most.

Code: Select all

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
aaargha
Filter Inserter
Filter Inserter
Posts: 333
Joined: Wed Dec 07, 2016 8:35 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by aaargha »

IT'S ALIVE!

The new test bench is go, updated OP with results, should contain all submitted intersections up to here (except Frodos, but that one needs a bit more work to become testable, and zytukins as there is no bp-string).

The save and mod should be up soon(TM).
User avatar
Dr. Walrus
Long Handed Inserter
Long Handed Inserter
Posts: 96
Joined: Fri Nov 20, 2015 6:30 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by Dr. Walrus »

I've got an intersection for you that allows U turns but won't deadlock.
Rank: D
Ouch.
gyro2death
Burner Inserter
Burner Inserter
Posts: 14
Joined: Mon Aug 29, 2016 3:14 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by gyro2death »

Ah thanks. Glad to see my compact spiral performance is pretty decent, I can now dare to say nothing is better at the same size according at least from what this thread shows. Oddly this thread just makes me want to go design even more now.... :D
gyro2death
Burner Inserter
Burner Inserter
Posts: 14
Joined: Mon Aug 29, 2016 3:14 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by gyro2death »

Okay new one to test, this is a symmetrical version of the "compact" one you've already tested. I'm wondering how it's throughput is compared and if it loses out. I assume they made it broke symmetry on purpose.

Code: Select all

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
Thank you very much for this thread!
aaargha
Filter Inserter
Filter Inserter
Posts: 333
Joined: Wed Dec 07, 2016 8:35 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by aaargha »

Dr. Walrus wrote:
I've got an intersection for you that allows U turns but won't deadlock.
Rank: D
Ouch.
Hehe :) Well, it is a pretty tricky deadlock so I wouldn't worry too much about it. I think it can be made safe if you move the entrances to the leaves further away from the central crossing, if it it's made possible to fit an entire train between the crossing and the entrance I believe it can no longer deadlock.
gyro2death wrote:Ah thanks. Glad to see my compact spiral performance is pretty decent, I can now dare to say nothing is better at the same size according at least from what this thread shows. Oddly this thread just makes me want to go design even more now.... :D
Excellent.... :D
gyro2death wrote:Okay new one to test, this is a symmetrical version of the "compact" one you've already tested. I'm wondering how it's throughput is compared and if it loses out. I assume they made it broke symmetry on purpose.

Thank you very much for this thread!
That one should be equivalent to the compact one. The reason that compact is not symmetrical is that with the closer rail spacing it would not be possible to separate the neighbouring input and output blocks. Take a look at "compact attempt" to see what happens.
Vals Loeder
Burner Inserter
Burner Inserter
Posts: 14
Joined: Wed May 24, 2017 11:58 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by Vals Loeder »

These are awesome. A great inspiration to my first experiments in rail in the game.
roman566
Fast Inserter
Fast Inserter
Posts: 137
Joined: Sat May 24, 2014 10:53 pm
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by roman566 »

I have a small problem and need help. I have created this intersection:

Code: Select all

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
It looks like this:
20170610113400_1.jpg
20170610113400_1.jpg (549.08 KiB) Viewed 24227 times
It's not visible on the screenshot, but the train that entered the intersection has stopped. The funny thing is that it doesn't happen all the time. In fact, it only seems to happen at random times when I am not looking. The question is - how do I fix it without throwing away all chain signals? I know it will work if I replace chain signals with normal signals, but it will reduce the amount of trains that can go through it at the same time to one. Which kind of kills the whole point of chain signals.
aaargha
Filter Inserter
Filter Inserter
Posts: 333
Joined: Wed Dec 07, 2016 8:35 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by aaargha »

roman566 wrote:It's not visible on the screenshot, but the train that entered the intersection has stopped. The funny thing is that it doesn't happen all the time. In fact, it only seems to happen at random times when I am not looking. The question is - how do I fix it without throwing away all chain signals? I know it will work if I replace chain signals with normal signals, but it will reduce the amount of trains that can go through it at the same time to one. Which kind of kills the whole point of chain signals.
What you are experiencing is this bug which is what makes roundabouts unsafe unless they're only one block. It's made even more prominent by the fact that there are two exits to the same direction from the same roundabout. This means that it will deadlock not only when a train decides to go a different direction part way through, but also if the train decides to take the other lane in the direction it was going. That kind of deadlock is very unlikely to resolve itself, it's impossible in many situations.

Replacing the chain signals with normal signals will cause the intersection to deadlock in heavy traffic, or just a few trains arriving at the same time if you're unlucky.

To be honest you may be better of, both stability and throughput wise, by going back to a 2-lane system with a good intersection.
roman566
Fast Inserter
Fast Inserter
Posts: 137
Joined: Sat May 24, 2014 10:53 pm
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by roman566 »

Thank you for that information. I will rebuild that one annoying intersection that caused me so much grief, but stick with single chunk roundabouts everywhere else, hoping that one day this bug will be fixed.
Stumpyofpain
Burner Inserter
Burner Inserter
Posts: 16
Joined: Tue Jun 06, 2017 4:31 pm
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by Stumpyofpain »

Hey

Could you test this 4 way crossing pls?
I plan to change my whole railways from 2 lane to 4 lane and for this i need a good cross :)
2017-06-10 20_49_35-Factorio 0.15.19.png
2017-06-10 20_49_35-Factorio 0.15.19.png (1.55 MiB) Viewed 24194 times
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
gyro2death
Burner Inserter
Burner Inserter
Posts: 14
Joined: Mon Aug 29, 2016 3:14 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by gyro2death »

aaargha wrote: That one should be equivalent to the compact one. The reason that compact is not symmetrical is that with the closer rail spacing it would not be possible to separate the neighbouring input and output blocks. Take a look at "compact attempt" to see what happens.
Ah! I knew I chose the right track spacing when I decided to have 3 tracks between them. I didn't even notice that one at first since I was only looking at the high throughput choices. Think that the compact is about as good as you can get for the space, but might try again and see if I can get the spiral to work with diagonal crossings if I make it larger. I like the look of it but can't afford the loss in throughput. Sadly, short of tearing down my base and rebuilding it with more space I likely won't be able to make a larger design work. Sadly didn't think about how many trains I would need once my needs for iron and copper got so high, takes a lot of trains to fill 10 blue belts worth of iron.
aaargha
Filter Inserter
Filter Inserter
Posts: 333
Joined: Wed Dec 07, 2016 8:35 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by aaargha »

Stumpyofpain wrote:Hey

Could you test this 4 way crossing pls?
I plan to change my whole railways from 2 lane to 4 lane and for this i need a good cross :)
Added as "Parentheses". Example of how it can deadlock is available in the album in the OP. Though that should be really rare so I wouldn't worry too much about it.
gyro2death wrote:Ah! I knew I chose the right track spacing when I decided to have 3 tracks between them. I didn't even notice that one at first since I was only looking at the high throughput choices. Think that the compact is about as good as you can get for the space, but might try again and see if I can get the spiral to work with diagonal crossings if I make it larger. I like the look of it but can't afford the loss in throughput. Sadly, short of tearing down my base and rebuilding it with more space I likely won't be able to make a larger design work. Sadly didn't think about how many trains I would need once my needs for iron and copper got so high, takes a lot of trains to fill 10 blue belts worth of iron.
That sound tricky indeed, if you've not already looked at it you can always consider longer trains and running them on rocket fuel to increase effective throughput.
StormCrow42
Burner Inserter
Burner Inserter
Posts: 14
Joined: Fri May 05, 2017 3:03 pm
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by StormCrow42 »

What really surprises me is the lower throughput on the 4-lane "simple" then on the Christmas design, where they seem almost identical pathwise. I'm guessing it's simply a factor of a smaller intersection size reduces the duration of blocked paths.
gyro2death
Burner Inserter
Burner Inserter
Posts: 14
Joined: Mon Aug 29, 2016 3:14 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by gyro2death »

SUCCESS! A TRUE COMPACT CELTIC KNOT!

Had to make its radius two rails longer than the absolute minimum and 1 rail longer then my own "Symmetrical Compact". However its still smaller then the compact you have listed in the main page and looks far better.

Looks like someone else is going to need to rename their 'Compact Celtic Knot' as I have truly made the compact version now! And if I'm right it should even have a small bit better throughput unless I screwed up a signal. Should be good for RHD and LHD though I could only test RHD on my setup. Anyways please and thank you for testing!

Code: Select all

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
Image this time! Don't mind the really awkward connection on the left, had to fit it into a spot where I was using my compact spiral which is 2 radius smaller. Image
aaargha
Filter Inserter
Filter Inserter
Posts: 333
Joined: Wed Dec 07, 2016 8:35 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by aaargha »

StormCrow42 wrote:What really surprises me is the lower throughput on the 4-lane "simple" then on the Christmas design, where they seem almost identical pathwise. I'm guessing it's simply a factor of a smaller intersection size reduces the duration of blocked paths.
The throughput loss is caused by the innermost diagonals being impossible to signal for a long stretch in the middle. This causes a lot of the paths to become dependant on each other, which leads to lower throughput.
gyro2death wrote:SUCCESS! A TRUE COMPACT CELTIC KNOT!

Had to make its radius two rails longer than the absolute minimum and 1 rail longer then my own "Symmetrical Compact". However its still smaller then the compact you have listed in the main page and looks far better.

Looks like someone else is going to need to rename their 'Compact Celtic Knot' as I have truly made the compact version now! And if I'm right it should even have a small bit better throughput unless I screwed up a signal. Should be good for RHD and LHD though I could only test RHD on my setup. Anyways please and thank you for testing!
Neat, I'll take a closer look in a while but I think there may actually be a way to deadlock that one with four trains. As for throughput, I don't think it'll differ much from the one it's based on but I'll give it a spin.
User avatar
Ristelle
Manual Inserter
Manual Inserter
Posts: 4
Joined: Sun Jun 04, 2017 8:14 am
Contact:

Re: 4-way intersection testing: Throughput and deadlocks

Post by Ristelle »

Coukd you test this roundabout in this blueprint book? cheers. https://pastebin.com/uEEtJMrL
Blah blah blah, Tidal-
Wait this isn't Tales of Vesperia? Well, drat!
Locked

Return to “Railway Setups”