3 and 4 way intersections

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.
Post Reply
Copy link
Meorin
Manual Inserter
Manual Inserter
Posts: 2
Joined: Wed Apr 23, 2025 5:27 pm
Contact:
Contact Meorin

Re: 3 and 4 way intersections

Post by Meorin »

Took me 2 hours of trial and error, just to end up at a cloverleaf interchange, but since it wasn't posted before, here we go:

Benefits: very few elevated rails and u-turn, exactly what I wanted for my first train setup.

LHT 82.43 size:86x94
Image

RHT 83.20 size:86x94
Image
Top
mmmPI
Smart Inserter
Smart Inserter
Posts: 4463
Joined: Mon Jun 20, 2016 6:10 pm
Contact:
Contact mmmPI

Re: 3 and 4 way intersections

Post by mmmPI »

Meorin wrote: Wed Apr 23, 2025 6:33 pm Took me 2 hours of trial and error, just to end up at a cloverleaf interchange, but since it wasn't posted before, here we go:

Benefits: very few elevated rails and u-turn, exactly what I wanted for my first train setup.
Unfortunatly, it look like the exact same as viewtopic.php?p=623353#p623353 which deadlock the test with current signaling.
Top
Meorin
Manual Inserter
Manual Inserter
Posts: 2
Joined: Wed Apr 23, 2025 5:27 pm
Contact:
Contact Meorin

Re: 3 and 4 way intersections

Post by Meorin »

mmmPI wrote: Wed Apr 23, 2025 7:26 pm
Meorin wrote: Wed Apr 23, 2025 6:33 pm Took me 2 hours of trial and error, just to end up at a cloverleaf interchange, but since it wasn't posted before, here we go:

Benefits: very few elevated rails and u-turn, exactly what I wanted for my first train setup.
Unfortunatly, it look like the exact same as viewtopic.php?p=623353#p623353 which deadlock the test with current signaling.
Damn it. Didn't see that one. Mine is very close but not the same. Got 8 more signals and with those extra signals it just went through a 1000min test without getting into a deadlock (LHT). You had me paranoid those 2 hours ended up in a flawed design but looks good to me still. Will test RHT for 1000min as well, just to be sure.

Image
Top
mmmPI
Smart Inserter
Smart Inserter
Posts: 4463
Joined: Mon Jun 20, 2016 6:10 pm
Contact:
Contact mmmPI

Re: 3 and 4 way intersections

Post by mmmPI »

Meorin wrote: Wed Apr 23, 2025 8:18 pm Damn it. Didn't see that one. Mine is very close but not the same. Got 8 more signals and with those extra signals it just went through a 1000min test without getting into a deadlock (LHT). You had me paranoid those 2 hours ended up in a flawed design but looks good to me still. Will test RHT for 1000min as well, just to be sure.
Maybe because there is only the set 1 ?
Top
Zavian
Smart Inserter
Smart Inserter
Posts: 1655
Joined: Thu Mar 02, 2017 2:57 am
Contact:
Contact Zavian

Re: 3 and 4 way intersections

Post by Zavian »

mmmPI wrote: Thu Apr 24, 2025 3:17 am Maybe because there is only the set 1 ?
I doubt that running only set one caused the lack of deadlocks. (I tested a couple of different LHD cloverleaf designs earlier today, and both deadlocked on set one within 90 minutes. But my designs didn't have all those extra signals).
Meorin wrote: Wed Apr 23, 2025 8:18 pm Got 8 more signals and with those extra signals it just went through a 1000min test without getting into a deadlock (LHT).
Here is what a potential deadlock situation looks like. (Handcrafted. It is possible that the timings in the testbed never actually line up to produce this particular situation. But theoretically it could happen).
Screenshot 2025-05-21 053540.png
Screenshot 2025-05-21 053540.png (881.16 KiB) Viewed 821 times
Top
mmmPI
Smart Inserter
Smart Inserter
Posts: 4463
Joined: Mon Jun 20, 2016 6:10 pm
Contact:
Contact mmmPI

Re: 3 and 4 way intersections

Post by mmmPI »

Zavian wrote: Tue May 20, 2025 7:45 pm
mmmPI wrote: Thu Apr 24, 2025 3:17 am Maybe because there is only the set 1 ?
I doubt that running only set one caused the lack of deadlocks. (I tested a couple of different LHD cloverleaf designs earlier today, and both deadlocked on set one within 90 minutes. But my designs didn't have all those extra signals).
It would be hard to believe that a full test doesn't yield a deadlock , or it would mean the test needs to be refined imo, because without chain signals and proper 2-4 trains my tests on linked bp resulted in deadlock all the time. So i thought maybe an old version of the test that was only done with trains coming from "some" direction that would be different for set 1 and 2 and 3 could have lead to a result with with only set 1 no deadlock occured.
Top
dryvnt
Manual Inserter
Manual Inserter
Posts: 2
Joined: Sat May 31, 2025 11:06 am
Contact:
Contact dryvnt

Re: 3 and 4 way intersections

Post by dryvnt »

It's obviously bad for longer trains, which this thread is clearly focused on, but I would like to at least document the humble roundabout. It's not megabase worthy, but I like to do smaller-scale grid-based stuff for easy supply chain management with 1-1 trains, so it's very useful to know about for someone like me.

After some testing and tweaking, I can conclude:
  • Works decently for 1-1 trains
  • Functional for 1-2 trains, but noticeably slower
  • Deadlocks on anything longer
  • Deadlocks can be worked around by turning all the inner signals into chain signals, at a cost of overall throughput
note: data+bp removed for being misleading, see post below
Last edited by dryvnt on Sun Jun 01, 2025 3:49 pm, edited 11 times in total.
Top
dryvnt
Manual Inserter
Manual Inserter
Posts: 2
Joined: Sat May 31, 2025 11:06 am
Contact:
Contact dryvnt

Re: 3 and 4 way intersections

Post by dryvnt »

Oh no, I seem to have had mods enabled (K2) that changed the results a bit. I apologize, that is my mistake for being inattentive. I have removed the misleading data from my post above. Here are some more accurate results for the humble roundabout:

1-1 score: 58.76 (90-min test)
1-1-1 score: 38.55 (90-min test)
1-3 score: 27.61 (modified signals, 30-min test)

Seeing how the results scale, I decided it wasn't worth investigating larger train sizes. The numbers are clear: The humble roundabout is obviously only worth considering for 1-1 trains, but specifically for 1-1 trains it performs very impressively for its footprint IMO.

06-01-2025, 17-36-04.png
06-01-2025, 17-36-04.png (970.42 KiB) Viewed 423 times
Top
Post Reply

Return to “Railway Setups”