Factorio Forums

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

3 and 4 way intersections

https://forums.factorio.com/viewtopic.php?f=194&t=100614

Page 4 of 15

Re: 3 and 4 way intersections

Posted: Sun Nov 28, 2021 2:38 am
by mrvn
Kano96 wrote: ↑
Sat Nov 27, 2021 8:54 pm
 Another related issue is also the question of rail signal placement. You can place signals in multiple different ways, like purely performance optimized or allowing trains to repath while waiting or like they would be in an actual rail set. All of these make some sense from a gameplay perspective but deliver different test results. Until now, we just let the author decide how to signal them for the most part, but we'll probably have to enforce one of the methods to keep test results comparable.
Could we add the rule that all images and blueprints must be attached instead of 3rd party services please.

Re: 3 and 4 way intersections

Posted: Sun Nov 28, 2021 6:35 am
by FuryoftheStars
mrvn wrote: ↑
Sun Nov 28, 2021 2:38 am
Kano96 wrote: ↑
Sat Nov 27, 2021 8:54 pm
 Another related issue is also the question of rail signal placement. You can place signals in multiple different ways, like purely performance optimized or allowing trains to repath while waiting or like they would be in an actual rail set. All of these make some sense from a gameplay perspective but deliver different test results. Until now, we just let the author decide how to signal them for the most part, but we'll probably have to enforce one of the methods to keep test results comparable.
Could we add the rule that all images and blueprints must be attached instead of 3rd party services please.
I don't care so much about attached vs 3rd party, but I would like to see images with all blueprints. Can't exactly see someone's intersection from just a blueprint when I'm not in a position to fire up the game....

Re: 3 and 4 way intersections

Posted: Sun Nov 28, 2021 8:59 pm
by mrvn
FuryoftheStars wrote: ↑
Sun Nov 28, 2021 6:35 am
mrvn wrote: ↑
Sun Nov 28, 2021 2:38 am
Kano96 wrote: ↑
Sat Nov 27, 2021 8:54 pm
 Another related issue is also the question of rail signal placement. You can place signals in multiple different ways, like purely performance optimized or allowing trains to repath while waiting or like they would be in an actual rail set. All of these make some sense from a gameplay perspective but deliver different test results. Until now, we just let the author decide how to signal them for the most part, but we'll probably have to enforce one of the methods to keep test results comparable.
Could we add the rule that all images and blueprints must be attached instead of 3rd party services please.
I don't care so much about attached vs 3rd party, but I would like to see images with all blueprints. Can't exactly see someone's intersection from just a blueprint when I'm not in a position to fire up the game....
+1

Re: 3 and 4 way intersections

Posted: Sun Nov 28, 2021 10:27 pm
by hansjoachim
mrvn wrote: ↑
Sun Nov 28, 2021 2:38 am
Could we add the rule that all images and blueprints must be attached instead of 3rd party services please.
There is a post character limit so we wouldn't have enough space for all intersections. We have backups so it's fine. If you mean replies i don't see the problem with factoriobin also i don't like setting rules that can't be enforced.
FuryoftheStars wrote: ↑
Sun Nov 28, 2021 6:35 am
I don't care so much about attached vs 3rd party, but I would like to see images with all blueprints. Can't exactly see someone's intersection from just a blueprint when I'm not in a position to fire up the game....
Yeah, that would be nice, but won't make it a requirement.

Re: 3 and 4 way intersections

Posted: Sun Nov 28, 2021 10:38 pm
by hansjoachim
mmmPI wrote: ↑
Sat Nov 27, 2021 9:47 pm
There is this other junction that was made after Koub made a thread about rail spacing. The goal was to make a 32x32 junction for it to be chunk-aligned. This is not my creation but it was the result that was choosen.(i think it was from hansjoachim)The rail spacing is 6 tile too, there was unsuccessful attempt to make it smaller by me. :). I didn't found it in the googledoc, it look like the "compact spiral" in the 2 lane unbuffered junctions from the csv/previous thread but with improved signaling it should score much better.
Yeah, i have been thinking about that intersection. Adding it:)

Re: 3 and 4 way intersections

Posted: Fri Dec 03, 2021 6:15 pm
by SaiMoen
I also had this one, managed to get 1:50, 2:39, 3:48, s:46 with optimal exit blocks.

Re: 3 and 4 way intersections

Posted: Sat Dec 04, 2021 12:02 am
by hansjoachim
SaiMoen wrote: ↑
Fri Dec 03, 2021 6:15 pm
 I also had this one, managed to get 1:50, 2:39, 3:48, s:46 with optimal exit blocks.

Thanks:)
That is the same one as already on the list. Crushed wide improved just RHD. I added it as an RHD version. The results are within margin of error.

Re: 3 and 4 way intersections

Posted: Thu Dec 09, 2021 7:32 pm
by ahopefulhobbit
Thanks to everyone who's added LHD blueprints! I vastly prefer LHD, but the old post didn't have many blueprints oriented that way

Re: 3 and 4 way intersections

Posted: Fri Dec 17, 2021 7:01 pm
by SaiMoen
Finally managed to match the score of the number 1 on the 2-lane 4-way unbuffered section :D
It took some tweaking, I tried to force it to be spaced by 2 tiles, but I could only make it work with spacing 4 (on RHD at least). I call it Snowball
1:50, 2:39, 3:51, s:47
Probably got lucky though, I feel like something is not quite right with a default parameter because I've seen an error of ± 2 tpm which seems a bit ridiculous.
Snowball.png
Snowball.png (669.64 KiB) Viewed 6253 times

Re: 3 and 4 way intersections

Posted: Sat Dec 18, 2021 12:33 am
by dinglebarry
Compact Celtic Knot with Narrow Tracks
Score: 46

Blueprint: LHD RHD Size: 38 x 38 Spacing: 2 tiles
LHD: Set 1: 50, Set 2: 38, Set 3: 49, Score: 46, RHD: Set 1: 50, Set 2: 39, Set 3: 49, Score: 46,
Author: dinglebarry
LHD pic.jpg
LHD pic.jpg (208.49 KiB) Viewed 6146 times
Inspiration taken from gyro2death's Super Compact Celtic Knot (#3 in 4-way unbuffered). I like the look and performance of the Celtic Knot intersection, but I also like having narrow tracks. So I started by using gyro2death's design, but narrowed the tracks and adjusted their positions to make space for signals. Then got rid of the unnecessary signals reducing the number from 36 to 24. It scores the same overall as gyro2death's intersection but mine performs slightly better on Set 1, while his performs slightly better on Set 3. Mine also has the advantage of being able to work just as well with RHD, while with his, he would need to have slightly less optimal signaling for RHD.

Re: 3 and 4 way intersections

Posted: Sat Dec 18, 2021 5:41 am
by hansjoachim
dinglebarry wrote: ↑
Sat Dec 18, 2021 12:33 am
 Compact Celtic Knot with Narrow Tracks
Score: 55

Blueprint: LHD RHD Size: 38 x 38 Spacing: 2 tiles
LHD: Set 1: 56, Set 2: 42, Set 3: 66, Score: 55, RHD: Set 1: 56, Set 2: 42, Set 3: 66, Score: 55
Author: dinglebarry
LHD pic.jpg

Inspiration taken from gyro2death's Super Compact Celtic Knot (#3 in 4-way unbuffered). I wanted to make a celtic knot intersection that would work with a narrower spacing between the tracks. So I started by using gyro2death's design, but narrowed the tracks and adjusted their positions to make space for signals. Then got rid of the unnecessary signals reducing the number from 36 to 24 and replaced some chain signals with normal signals to speed up the intersection. To my surprise it ended up having a significantly higher throughput than the original design in all sets. To be fair, it felt like I was cheating in Set 3 since it was constantly prioritizing the trains moving vertically and it never switched to allow the trains moving horizontally to get through.
Thanks for the submission:) looks like a good improvement but it seems you didn't test with safe outputs and that is why your results are so high.

Re: 3 and 4 way intersections

Posted: Sat Dec 18, 2021 6:55 pm
by Hovedgade
I'm back once again with a new roundabout. :D
This one is smaller than hansjoachim's Parallel split medium but has the same score and the same spacing. Without the circuitry the throughput of the intersection will fall drasticly. I tested it with unsafe outputs/exits because this intersections has those bufferzones included. I garentee that this intersection will not deadlock.
Flower roundabout
Score: 100

Size: 230x230, Spacing: 6 tiles, Train Length: 6
RHD: Set1: 103, Set2: 103, Set3: 95
Author: Hovedgade
Image
Edit: By accident i wrote the wrong dimensions last time. It should be fixed now

Re: 3 and 4 way intersections

Posted: Sun Dec 19, 2021 12:11 am
by dinglebarry
hansjoachim wrote: ↑
Sat Dec 18, 2021 5:41 am
dinglebarry wrote: ↑
Sat Dec 18, 2021 12:33 am
 Compact Celtic Knot with Narrow Tracks
Score: 55

Blueprint: LHD RHD Size: 38 x 38 Spacing: 2 tiles
LHD: Set 1: 56, Set 2: 42, Set 3: 66, Score: 55, RHD: Set 1: 56, Set 2: 42, Set 3: 66, Score: 55
Author: dinglebarry
LHD pic.jpg

Inspiration taken from gyro2death's Super Compact Celtic Knot (#3 in 4-way unbuffered). I wanted to make a celtic knot intersection that would work with a narrower spacing between the tracks. So I started by using gyro2death's design, but narrowed the tracks and adjusted their positions to make space for signals. Then got rid of the unnecessary signals reducing the number from 36 to 24 and replaced some chain signals with normal signals to speed up the intersection. To my surprise it ended up having a significantly higher throughput than the original design in all sets. To be fair, it felt like I was cheating in Set 3 since it was constantly prioritizing the trains moving vertically and it never switched to allow the trains moving horizontally to get through.
Thanks for the submission:) looks like a good improvement but it seems you didn't test with safe outputs and that is why your results are so high.
Oh, I honestly don't know what safe outputs means, but I am guessing that it means I would need to have enough space in the exit block to fit an entire train. I can try to do that and retest it.

EDIT: Nevermind. I went back to the original post in the forum and figured out what you were talking about. I don't know how I missed that the first time. Thanks for the help though. I really couldn't understand why my score was significantly higher than the other intersections in that category. I edited the first post so now it has the correct scores.

Re: 3 and 4 way intersections

Posted: Tue Dec 21, 2021 11:07 pm
by hansjoachim
Happy holidays:)
I will go over the submitted intersections eventually, it's just holidays:)

Cross Parallel Buffer

Posted: Sun Dec 26, 2021 9:00 am
by Avona
I made a parallel buffered version of the cross intersection!
Cross Parallel Buffer
Score 82

RHD
Set 1: 100, 97
Set 2: 80, 81
Set 3: 68, 68

LHD
Set 1: 99
Set 2: 79
Set 3: 68

6 car trains, 174x174, 2 lane, 6 tile spacing



Image

Starfish

Posted: Sun Dec 26, 2021 1:51 pm
by Avona
Starfish v1.2
Edits were made so that set 3 now alternates properly. Signals were reduced for ups concerns.

Score 94

RHD
S1: 104
S2: 91
S3: 87

LHD
S1: 103
S2: 92
S3: 89

6 car trains, 150x150, 2 lane, 6 tile spacing



Image

Hurricane Intersection

Posted: Mon Dec 27, 2021 10:31 am
by Avona
After more adjustments, I present Hurricane 1.1.

6 car trains, 150x150, 2 lane, 6 tile spacing

Score 108

RHD
S1: 107, 107, 108
S2: 106, 106, 106
S3: 111, 111, 111

LHD
S1: 109
S2: 105
S3: 110



Image

Chubby Bunny

Posted: Thu Dec 30, 2021 12:54 pm
by Avona
I found that during set 2, only one set of lanes would be utilized, reducing overall throughput. There isn't enough space to modify track lengths so I put a circuit network connection on the signals that only allows trains from a direction to go if they have the same or more buffered trains than the compared lane. I'd rather not use circuits but, after hours of testing today, it seems that these circumstances might actually happen in an operating factory, so I went with the circuit conditions.

I've included the circuited versions and uncircuited versions now.

Chubby Bunny Circuited
Score 77

RHD
S1: 81
S2: 72

LHD
S1: 80
S2: 73

Uncircuited
Score 57

RHD
S1: 79
S2: 34

LHD
S1: 80
S2: 34

6 car trains, 150x100, 2 lane, 6 tile spacing



Image

Echo 1.1

Posted: Fri Dec 31, 2021 4:42 am
by Avona
Echo is another intersection based on the cross parallel buffer which is a bit squashed but retains its mirror symmetry.
Edit: I modified Echo so that set 3 now synchronizes. The stat line is now significantly better than the cross parallel buffer.
I've also just added the Echo 1.1 3 Way! It converts to the echo easily! (I was unable to synchronize the 3 Way LHD. I will keep trying.)

Echo 1.1 4 Way 150x150

RHD
Score 89
S1: 101
S2: 80
S3: 86

LHD
Score 90
S1: 102
S2: 83
S3: 85

Echo 1.1 3 Way 150x112

RHD
Score 62
S1: 78
S2: 46

LHD
Score 56
S1: 78
S2: 34

6 car trains, 2 lane, 6 tile spacing



Image

Image

Baby Starfish

Posted: Sat Jan 01, 2022 10:04 am
by Avona
This is based on the starfish but not parallel buffered. I was going for minimum width while keeping buffers.

Baby Starfish 1.1 (Now with less signals!)

RHD
Score 59
S1: 76
S2: 51
S3: 50

LHD
Score 59
S1: 75
S2: 51
S3: 50

6 car trains, 90x90, 2 lane, 6 tile spacing



Image
All times are UTC
Page 4 of 15

Powered by phpBB® Forum Software © phpBB Limited