Page 21 of 24

Re: 3 and 4 way intersections

Posted: Mon Nov 11, 2024 11:18 pm
by coppercoil
Factoriointersection wrote: Mon Nov 11, 2024 7:26 am Yes, I can rename. Both the intersection you posted deadlock. For intersections with U-turns, they should be tested with the Everything test. Actually so does the first one you posted, so I'll take them off the site untill they don't deadlock.
I fixed my intersections. I updated images and blueprints in my original posts:
Lilac+U viewtopic.php?p=629757#p629757
Bottle+U, Airplane+U viewtopic.php?p=636998#p636998

Re: 3 and 4 way intersections

Posted: Tue Nov 12, 2024 4:55 am
by sparr
My attempt at a nice simple one chunk one layer intersection with four way rotational symmetry and LHD. RHD is easier, see the next post in the thread for the simple solution to that.
32x32 tiles
2-rail spacing
Score: 44.24

One (well, four) rail signal block is larger than I'd like, but not in a conflicting way. I think this is the minimum size for an intersection in 2.0? I suspect someone has posted this somewhere already, but I couldn't find it at the top of either thread.

If I have to name it... "Small Askew"

Also, another metric I think we should care about... The shortest exit block section is ~3 rails (higher number is better, it means the exit block on the main rail can be 3 tiles shorter than if there was a rail signal just beyond the edge of this blueprint).
11-11-2024, 23-45-36.png
11-11-2024, 23-45-36.png (525.39 KiB) Viewed 2765 times

Re: 3 and 4 way intersections

Posted: Tue Nov 12, 2024 5:24 pm
by TimEv
Surprised that I haven't seen my chunk aligned 4 way intersection.
11-12-2024, 18-21-43.png
11-12-2024, 18-21-43.png (958.41 KiB) Viewed 2707 times
11-12-2024, 18-19-15.png
11-12-2024, 18-19-15.png (37.93 KiB) Viewed 2707 times


Also: first time posting here, hope I'm doing it right!

EDIT: well seems i wasn't doing it right, i looked in the 4 lane in stead of 2 lane for some reason :(. The score is still slightly better than Avona's symmetrical cross, using the exact same rail layout, just slightly differently placed signals. Is there some variance in the testing or something?

Re: 3 and 4 way intersections

Posted: Wed Nov 13, 2024 5:17 am
by Avona
TimEv wrote: Tue Nov 12, 2024 5:24 pm The score is still slightly better than Avona's symmetrical cross, using the exact same rail layout, just slightly differently placed signals. Is there some variance in the testing or something?
It could be a variety of things. Signal placement could have some effect on how the trains interact. I also notice that you used a 30 minute test, which will have more variance. The tests are random, so there is quite a bit of variance, especially with low test times.
From the testbenchcontrols faq:
"Q: What is the margin of errror in testresults?
A: roughly for quick tests 30min, it's about +- 1. For medium tests its +-0,5 for long test its +- 0.1."

Re: 3 and 4 way intersections

Posted: Wed Nov 13, 2024 11:59 pm
by Hovedgade
A buffered turbo roundabout for those who like (and don't like) roundabouts

Small Turbo Roundabout
RHT, 4 tile spacing, 116x116 footprint, made for 6 wagon trains
Image

:D

Re: 3 and 4 way intersections

Posted: Fri Nov 15, 2024 1:47 pm
by zaanzabar
PiggyWhiskey wrote: Mon Nov 11, 2024 6:30 am I made a LHD version of the Elevated 2-Lane 4-Way by Zaanza

I eliminated to lane change which isn't needed as trains change earlier, and I shortened the lead in as there are long enough lengths within the intersection to hold any trains for the merge out.

11-11-2024, 17-29-36.png

11-11-2024, 17-28-14.png

Wow! Impressive results. Both things you changed, I had for my personal reasons. Thanks for making it universal.

Re: 3 and 4 way intersections

Posted: Fri Nov 15, 2024 9:23 pm
by Factoriointersection
Updated it.
There was an issue with the safeoutputs so I had to wait to test until I adjusted them and I have retested the 4 way 2 lane oneplane intersections.
Great new intersections added=) Awesome guys!
Hovedgade, nice to see you back!

Re: 3 and 4 way intersections

Posted: Fri Nov 15, 2024 11:26 pm
by Hovedgade
Factoriointersection wrote: Fri Nov 15, 2024 9:23 pm Hovedgade, nice to see you back!
thanks :)

I got some additional variations of the small turbo roundabout and It is easily possible to make them more compact but then they wouldn't be as beautiful. (One of them has traffic for the left side of the rails and a zero tile gap, so good luck benchmarking that one :twisted: )

Re: 3 and 4 way intersections

Posted: Sat Nov 16, 2024 2:27 pm
by Factoriointersection
coppercoil wrote: Mon Nov 11, 2024 11:18 pm
Factoriointersection wrote: Mon Nov 11, 2024 7:26 am Yes, I can rename. Both the intersection you posted deadlock. For intersections with U-turns, they should be tested with the Everything test. Actually so does the first one you posted, so I'll take them off the site untill they don't deadlock.
I fixed my intersections. I updated images and blueprints in my original posts:
Lilac+U viewtopic.php?p=629757#p629757
Bottle+U, Airplane+U viewtopic.php?p=636998#p636998
Great:) U turns are nice to include, maybe we should also show the score in everything test besides it? Added them.

Re: 3 and 4 way intersections

Posted: Sat Nov 16, 2024 2:28 pm
by Factoriointersection
Hovedgade wrote: Fri Nov 15, 2024 11:26 pm
Factoriointersection wrote: Fri Nov 15, 2024 9:23 pm Hovedgade, nice to see you back!
thanks :)

I got some additional variations of the small turbo roundabout and It is easily possible to make them more compact but then they wouldn't be as beautiful. (One of them has traffic for the left side of the rails and a zero tile gap, so good luck benchmarking that one :twisted: )
Nice :) many prefer design over throughput:)

Re: 3 and 4 way intersections

Posted: Sat Nov 16, 2024 4:21 pm
by coppercoil
Factoriointersection wrote: Sat Nov 16, 2024 2:27 pm maybe we should also show the score in everything test besides it?
I think, no :). If trains use U-turns frequently, it means there is something wrong with a railway. U-turns should be a rare exception, this is why I didn't try to make them fast. Of course, there should not be any deadlocks anyway.
If players need fast U-turns for a good reason, they may use some specialized node with no intersection, just the turns. Making fast U-turn in a regular intersection is more difficult, it makes the intersection larger and slower.

Re: 3 and 4 way intersections

Posted: Sun Nov 17, 2024 2:04 am
by Robert3D
Ive been trying to design some intersections for my SA playthrough. They use a 2-4-2 spacing because i like the symmetry and look. Ive been trying to balance compactness and throughput so while these arnt the fastest, they are much smaller then other designs. Both are LHT


Compact 3 Way

128 TPM
Size 84 x 58

3way.png
3way.png (1.11 MiB) Viewed 2235 times


Compact 4 Way

106 TPM
Size 104 x 92

4way.png
4way.png (1.38 MiB) Viewed 2235 times


The 4 Way technically allows for U-turns, but these should remain the exception. Too many trains trying to U-turn and turn right at the same time will dead lock it.

Both of these can probably still be improved but im satisfied with the results.
If you got any advice on how to make them better, let me know.

Re: 3 and 4 way intersections

Posted: Sun Nov 17, 2024 3:11 pm
by Factoriointersection
Robert3D wrote: Sun Nov 17, 2024 2:04 am Ive been trying to design some intersections for my SA playthrough. They use a 2-4-2 spacing because i like the symmetry and look. Ive been trying to balance compactness and throughput so while these arnt the fastest, they are much smaller then other designs. Both are LHT


Compact 3 Way

128 TPM
Size 84 x 58


3way.png


Compact 4 Way

106 TPM
Size 104 x 92


4way.png


The 4 Way technically allows for U-turns, but these should remain the exception. Too many trains trying to U-turn and turn right at the same time will dead lock it.

Both of these can probably still be improved but im satisfied with the results.
If you got any advice on how to make them better, let me know.
Hey, I added the 3 way=)
Since the 4 way deadlock I wont add it.

If the trains on the loop had a bit more space it would be better. If you make the blue areas larger so that the trains can go in the green directions that would help a lot.
11-17-2024, 16-11-28.png
11-17-2024, 16-11-28.png (1.16 MiB) Viewed 2167 times

Re: 3 and 4 way intersections

Posted: Sun Nov 17, 2024 3:33 pm
by Robert3D
Thanks, ill test around. The 4 way only deadlocks if a train attempts to U-turn, traffic in any other direction should never deadlock. In order to prevent that deadlock, i would probably need to kill the waiting area in the middle of the crossing, slowing down trains quite a bit but ill check.

Re: 3 and 4 way intersections

Posted: Sun Nov 17, 2024 5:07 pm
by Factoriointersection
Robert3D wrote: Sun Nov 17, 2024 3:33 pm Thanks, ill test around. The 4 way only deadlocks if a train attempts to U-turn, traffic in any other direction should never deadlock. In order to prevent that deadlock, i would probably need to kill the waiting area in the middle of the crossing, slowing down trains quite a bit but ill check.
Yeah. I have a test that uses the U-turn. Click the green settings button and pick everything test, then you can test U-turns and lane changes

Re: 3 and 4 way intersections

Posted: Wed Nov 20, 2024 6:57 pm
by koelkastdeur16
Rotationally symmetrical 3-chunck 2-lane RHD

Entrances are widely spaced, so i could fit my straight sections around my powerpole and roboport grid.
11-20-2024, 19-54-10.png
11-20-2024, 19-54-10.png (633.7 KiB) Viewed 1879 times

Re: 3 and 4 way intersections

Posted: Wed Nov 20, 2024 10:45 pm
by Factoriointersection
koelkastdeur16 wrote: Wed Nov 20, 2024 6:57 pm Rotationally symmetrical 3-chunck 2-lane RHD

Entrances are widely spaced, so i could fit my straight sections around my powerpole and roboport grid.

11-20-2024, 19-54-10.png

I like the shape:) I feel like the ramps destroyed the symmetry, nice to see it return:)

Re: 3 and 4 way intersections

Posted: Thu Nov 21, 2024 4:05 am
by Robert3D
Compact 4 Way v.2

150 TPM
Size 100 x 112
4way.png
4way.png (1.16 MiB) Viewed 1790 times

Did an "everything" test for 1000mins and it did not deadlock anymore.

Spamming rail signals really helped the rates too. Did anyone do a test on how much lag rail signals, small logic networks for rail signals and trains trying to path through excessive rail signal spam cause ? It seems beneficial to use as many rail signals as possible to boost rates, but it would be devastating for large builds and mega bases if it caused significant lag.

Re: 3 and 4 way intersections

Posted: Thu Nov 21, 2024 5:50 am
by Factoriointersection
Robert3D wrote: Thu Nov 21, 2024 4:05 am Compact 4 Way v.2

150 TPM
Size 100 x 112

4way.png

Did an "everything" test for 1000mins and it did not deadlock anymore.

Spamming rail signals really helped the rates too. Did anyone do a test on how much lag rail signals, small logic networks for rail signals and trains trying to path through excessive rail signal spam cause ? It seems beneficial to use as many rail signals as possible to boost rates, but it would be devastating for large builds and mega bases if it caused significant lag.
Awesome, looks good:)
It looks like the spamming isnt too bad. If you go signals closer than a wagons distance that can start costing a bit.

Re: 3 and 4 way intersections

Posted: Fri Nov 22, 2024 2:32 pm
by hooray
Robert3D wrote: Sun Nov 17, 2024 2:04 am Ive been trying to design some intersections for my SA playthrough. They use a 2-4-2 spacing because i like the symmetry and look. Ive been trying to balance compactness and throughput so while these arnt the fastest, they are much smaller then other designs. Both are LHT


Compact 3 Way

128 TPM
Size 84 x 58


3way.png


Compact 4 Way

106 TPM
Size 104 x 92


4way.png


The 4 Way technically allows for U-turns, but these should remain the exception. Too many trains trying to U-turn and turn right at the same time will dead lock it.

Both of these can probably still be improved but im satisfied with the results.
If you got any advice on how to make them better, let me know.
i need RHT sir :cry: PLZ