Factorio Forums

www.factorio.com
https://forums.factorio.com/

3 and 4 way intersections

https://forums.factorio.com/viewtopic.php?t=100614

Page 17 of 25

Re: 3 and 4 way intersections

Posted: Sat Oct 26, 2024 12:40 pm
by TBTerra2
im unsure how well the tester will deal with these, but i find them very useful in mega bases
4 lain mainline with a 2lane branch off it. the pathing weights have to be modified abit to get it working right (should have about equal spawns/outputs for each seperate direction, and 4lane parts should have about twice the spawns as 2lane)
LHD
RHD
10-26-2024, 13-39-03.png
10-26-2024, 13-39-03.png (436.65 KiB) Viewed 3471 times

Re: 3 and 4 way intersections

Posted: Sat Oct 26, 2024 7:00 pm
by hansjoachim
TBTerra2 wrote: Sat Oct 26, 2024 12:40 pm im unsure how well the tester will deal with these, but i find them very useful in mega bases
4 lain mainline with a 2lane branch off it. the pathing weights have to be modified abit to get it working right (should have about equal spawns/outputs for each seperate direction, and 4lane parts should have about twice the spawns as 2lane)
LHD
RHD
10-26-2024, 13-39-03.png
The everything test handles it fine and is pretty balanced. Spacing got messed up though XD Have to make a seperate catagory for this one=)
10-26-2024, 21-00-11.png
10-26-2024, 21-00-11.png (28.16 KiB) Viewed 3408 times

Re: 3 and 4 way intersections

Posted: Sat Oct 26, 2024 9:06 pm
by Bocian

Test results from RailTester in BP descriptions. Sorry I cut off the decimals... Everything is rounded down unless it was something like .98 where I'd round up. Settings were as follows, except I tested for a bunch of train configs:
10-26-2024, 22-59-28.png
10-26-2024, 22-59-28.png (22 KiB) Viewed 3366 times
All of these intersections are optimised primarily around 6 car trains, and secondarily 3 and 2 cars. All RHD.
All designs aside from Paperclip are mine made before SA with Fake New Rails first, now recreated with real ones. Paperclip was posted in LHD version on Factorio Discord by Xortle - I just made an RHD version and named it.
10-26-2024, 22-56-02.png
10-26-2024, 22-56-02.png (490.8 KiB) Viewed 3366 times
10-26-2024, 22-56-14.png
10-26-2024, 22-56-14.png (549.28 KiB) Viewed 3366 times
10-26-2024, 22-56-27.png
10-26-2024, 22-56-27.png (563.08 KiB) Viewed 3366 times
10-26-2024, 22-56-37.png
10-26-2024, 22-56-37.png (685.15 KiB) Viewed 3366 times
10-26-2024, 22-56-57.png
10-26-2024, 22-56-57.png (621.1 KiB) Viewed 3366 times
10-26-2024, 22-57-09.png
10-26-2024, 22-57-09.png (574.07 KiB) Viewed 3366 times
10-26-2024, 22-57-20.png
10-26-2024, 22-57-20.png (599.92 KiB) Viewed 3366 times
10-26-2024, 22-57-30.png
10-26-2024, 22-57-30.png (646.95 KiB) Viewed 3366 times
10-26-2024, 22-57-40.png
10-26-2024, 22-57-40.png (1.14 MiB) Viewed 3366 times

Re: 3 and 4 way intersections

Posted: Sat Oct 26, 2024 9:17 pm
by hansjoachim
[/quote]
Tried testbenchcontrols 2.0.4 but it still detects this intersection as LHT
[/quote]

Now i re did the 1.1 to 2.0 conversion and LHT/RHT should be fixed, including also the Everything test and also a couple other things:)

Re: 3 and 4 way intersections

Posted: Sun Oct 27, 2024 1:38 am
by mmmPI
Reverse Manual Nightmare

RHT- 4 tile spacing between rails
Size : 96x96 = 3 chunks

If you drive horizontal, you can go manual, but if you go vertical, no, it is all in reverse, hence the name.
Reverse Manual Nightmare.jpg
Reverse Manual Nightmare.jpg (170.69 KiB) Viewed 3283 times

Re: 3 and 4 way intersections

Posted: Mon Oct 28, 2024 3:41 am
by lucyisgamer
Chunky Mk. 4
RHT - 4 tile spacing
Size - 96x96 (3x3 chunks)

Originally built for 2-4 trains for easy comparison with the ones in the list, I've also tested with 2-8 trains, as that's what I'll end up using.
Test Results
Size | Set 1 | Set 2 | Set 3 | Score
2 - 4 | 101.79 | 106.42 | 113.62 | 107.28
2 - 8 | 64.53 | 66.60 | 84.87 | 72.00
blueprint.PNG
blueprint.PNG (3.3 MiB) Viewed 3091 times
I might take some time and make a LHT version of this, we'll see :)

Re: 3 and 4 way intersections

Posted: Mon Oct 28, 2024 3:58 am
by OzarkRanger
MJDSys wrote: Thu Oct 24, 2024 8:57 pm I played around with a bowed 4 way cross after seeing some other examples and it has slightly higher performance. It's also single chunk aligned:
10-24-2024, 16-45-59.png
10-24-2024, 16-46-20.png
I was working on something similar and ended up with a symmetrical version of the same. It still fits into a 32x32 chunk, but the performance is slightly worse for some reason.
Screenshot 2024-10-27 220639.png
Screenshot 2024-10-27 220639.png (1.06 MiB) Viewed 3090 times


This is my first time using the testbench and I couldn't get it run more than one set for some reason. If someone who knows what they're doing can retest this, I would appreciate it.
Screenshot 2024-10-27 214317.png
Screenshot 2024-10-27 214317.png (30.27 KiB) Viewed 3090 times

It also expands nicely from the three-way equivalent:
Screenshot 2024-10-27 220548.png
Screenshot 2024-10-27 220548.png (688.74 KiB) Viewed 3090 times


Interestingly, this only uses 56 pieces of rail compared to 68 for the straight version, which makes it very tempting to switch entirely to rick-rack train lines. :)
Screenshot 2024-10-27 224915 adj.png
Screenshot 2024-10-27 224915 adj.png (1.72 MiB) Viewed 3090 times

Re: 3 and 4 way intersections

Posted: Mon Oct 28, 2024 6:45 am
by Avona
grossws wrote: Fri Oct 25, 2024 11:03 pm Tried testing newly designed intersections and I got slightly different results on my naive intersection updated for 2.0 (one plane) than similar looking Avona's "Symmetrical Cross". Likely because of a bit different signalling on right turns.
It's more than the right turns, it's the fact that I signaled all of the merges and splits differently. You don't need chains before merges or splits (unless it's a merge directly into a split). And the extra space that the chain blocks take up will reduce the throughput of your intersection significantly.

Re: 3 and 4 way intersections

Posted: Mon Oct 28, 2024 11:57 am
by akulen
I made a probably overkill 4 lane 4 way intersection. The blueprint includes exchangers as the right lane can't turn left (RHT).
It is signaled for 2-4 trains, but there is enough room to signal it for bigger trains, though the exchangers would need to be extended.
It was designed to be exactly 9 chunks wide, and rotationally symmetric.

Blueprint: RHT, Size: 288 x 288, Spacing: 4 tiles, Train length: 2-4
RHT: Set1: 136.73, Set2: 143.37, Set3: 147.87


4l4wITest.png
4l4wITest.png (33.82 KiB) Viewed 3022 times
4l4wIScreenshot.png
4l4wIScreenshot.png (446.27 KiB) Viewed 3022 times

Re: 3 and 4 way intersections

Posted: Mon Oct 28, 2024 6:04 pm
by Factoriointersection
Awesome guys! some really nice intersections! I added almost all. And updated the forum. Lots of intersection on there now=)

Re: 3 and 4 way intersections

Posted: Mon Oct 28, 2024 9:36 pm
by grossws
Avona wrote: Mon Oct 28, 2024 6:45 am
grossws wrote: Fri Oct 25, 2024 11:03 pm Tried testing newly designed intersections and I got slightly different results on my naive intersection updated for 2.0 (one plane) than similar looking Avona's "Symmetrical Cross". Likely because of a bit different signalling on right turns.
It's more than the right turns, it's the fact that I signaled all of the merges and splits differently. You don't need chains before merges or splits (unless it's a merge directly into a split). And the extra space that the chain blocks take up will reduce the throughput of your intersection significantly.
I know I technically don't need chain signals there thought it's a bit of habit. I was surprised that my variant I thought suboptimal somehow shows better throughput while it shouldn't. I'll test your variant later to have comparison in the same environment

Re: 3 and 4 way intersections

Posted: Tue Oct 29, 2024 11:47 am
by mmmPI
3 Chunk Square
3-chunk-square.jpg
3-chunk-square.jpg (181.2 KiB) Viewed 2792 times


I made this junction trying to optimize for the test bench, but i noticed the test bench adds some rails that invalidate the signals, if removed manually , the junction seem to score good , but i don't want to publish my own non-official results.

Re: 3 and 4 way intersections

Posted: Tue Oct 29, 2024 2:32 pm
by Factoriointersection
mmmPI wrote: Tue Oct 29, 2024 11:47 am
3 Chunk Square
3-chunk-square.jpg



I made this junction trying to optimize for the test bench, but i noticed the test bench adds some rails that invalidate the signals, if removed manually , the junction seem to score good , but i don't want to publish my own non-official results.
Looks really nice:)
I'll see if I can do something with the rail placing.

Re: 3 and 4 way intersections

Posted: Tue Oct 29, 2024 6:18 pm
by ManDeJan
Hi!

I'm new here, I made this 4 way intersection for my space age playthrough, I use 1-2-1 trains but for this thread I tested it with 2-4 trains :). I'm new to trains (previous playthroughs only did belts), this is the first intersection I designed, not sure all the signalling is exactly correct but it seems to run smoothly through the benchmark. I'm happy it fits exactly in a 5 tile cross.


Screenshot 2024-10-29 191343.png
Screenshot 2024-10-29 191343.png (1.43 MiB) Viewed 2718 times

Re: 3 and 4 way intersections

Posted: Tue Oct 29, 2024 7:32 pm
by Factoriointersection
mmmPI wrote: Tue Oct 29, 2024 11:47 am
3 Chunk Square
3-chunk-square.jpg



I made this junction trying to optimize for the test bench, but i noticed the test bench adds some rails that invalidate the signals, if removed manually , the junction seem to score good , but i don't want to publish my own non-official results.
I figured it out. It's because under the elevated rai there is an open rail. No Idea how I can account for that. I fixed it so that a singel rail wouldn't cause issues, but an open ended rail like that, hard to patch.
10-29-2024, 20-31-27.png
10-29-2024, 20-31-27.png (356.21 KiB) Viewed 2686 times

Re: 3 and 4 way intersections

Posted: Tue Oct 29, 2024 7:35 pm
by Factoriointersection
ManDeJan wrote: Tue Oct 29, 2024 6:18 pm Hi!

I'm new here, I made this 4 way intersection for my space age playthrough, I use 1-2-1 trains but for this thread I tested it with 2-4 trains :). I'm new to trains (previous playthroughs only did belts), this is the first intersection I designed, not sure all the signalling is exactly correct but it seems to run smoothly through the benchmark. I'm happy it fits exactly in a 5 tile cross.



Screenshot 2024-10-29 191343.png
Nice=) looks like about score 100 ish. You sure about the name though?

Re: 3 and 4 way intersections

Posted: Tue Oct 29, 2024 8:14 pm
by Locutus123456
Hi All,

I posted an intersection on Reddit (https://www.reddit.com/r/factorio/comme ... ersection/) and was pointed to this forum.

To be honest at first I did not see the added value of the benchmarking mod. The reason being that this junction does not actually contain same level intersections and by first splitting off traffic that goes in different directions I minimize conflict points. Only if two trains share the same outbound direction they will have to wait for each other. So I, incorrectly reasoned: "what is left to test".

However, running the benchmarking tool I was a bit disappointment in the performance. First, because this signal density is much higher than what I'm used to when playing Factorio, but this was easily fixed. Secondly, this test contains, many trains that simultaneously want to go in the same direction, this will quickly create a traffic jam on the single inbound track.

I then started experimenting with buffering the outbound traffic, as you can see below there is quite some space before the outbound traffic is merged into a single track. It does however feel like cheating a bit, how much buffer can you add while still considering it as two lane traffic?
Screenshot 2024-10-29 204545.png
Screenshot 2024-10-29 204545.png (436.73 KiB) Viewed 2445 times


Final note: I'm aware that my original design without the buffer lanes has been posted before, I was unaware of this prior work.

Re: 3 and 4 way intersections

Posted: Wed Oct 30, 2024 6:24 am
by akulen
I finally improved my previous intersection, adding the last pair of direction/lane pair that was missing.
It can probably be improved on, but at least it makes for a reasonable 4 lane 4 way intersection baseline.
It fits exactly in 6x6 chunks

Blueprint: RHT, Size: 192 x 192, Spacing: 4 tiles, Train length: 2-4
RHT: Set1: 174.09, Set2: 188.80, Set3: 197.20


StarLoopTest.png
StarLoopTest.png (35.02 KiB) Viewed 2581 times
StarLoopScreenshot.png
StarLoopScreenshot.png (990.91 KiB) Viewed 2581 times

Re: 3 and 4 way intersections

Posted: Wed Oct 30, 2024 8:23 pm
by Factoriointersection
akulen wrote: Wed Oct 30, 2024 6:24 am I finally improved my previous intersection, adding the last pair of direction/lane pair that was missing.
It can probably be improved on, but at least it makes for a reasonable 4 lane 4 way intersection baseline.
It fits exactly in 6x6 chunks

Blueprint: RHT, Size: 192 x 192, Spacing: 4 tiles, Train length: 2-4
RHT: Set1: 174.09, Set2: 188.80, Set3: 197.20



StarLoopTest.png
StarLoopScreenshot.png
nice one! added it. You should update and use the latest scenario=)

Re: 3 and 4 way intersections

Posted: Thu Oct 31, 2024 6:48 am
by Avona
Frog Eye Intersection: 6 tile (3 spaced), 94x86
https://factoriobin.com/post/wg4eu1
Image
Image
All times are UTC
Page 17 of 25

Powered by phpBB® Forum Software © phpBB Limited