that train going to an occupied slot is an old problem.
one solution is to have a unique named station before the others.
have the train stop for 0 seconds, it will then recalculate to the next stop.
i'm sure there are others. i'm working on one right now, but not ready to release it yet.
hub/"entrepot" based factory; station designs?
-
impetus maximus
- Smart Inserter
- Posts: 1299
- Joined: Sat Aug 20, 2016 10:07 pm
- Contact:
-
vanatteveldt
- Filter Inserter
- Posts: 947
- Joined: Wed Nov 25, 2015 11:44 am
- Contact:
Re: hub/"entrepot" based factory; station designs?
I tried that solution, but it added such a delay that the waiting bays were actually not used with 4 stops and a fully saturated input line.
289 seconds for 50 trains with pre-stop, 238 seconds for 50 trains without the stop.
Edit: nm, there was a signalling issue. With 8 stops time is just under 200 seconds for 50 trains, adding a temporary stop does not seem to change much. The empty bay syndrom seems to be solved now there is enough room for a full train after the junction plus a signal just before
Test setup:
I added an extra track to the left to bypass waiting trains for trains exiting S (this solves the loop/deadlock problem), and made it a bit less compact to ensure a full train can wait between junction and first loading bay.
This is slighty better than the loop stations (a train every 4.5 seconds rather than every 6). Adding some extra signals gets it further down (to around 170-180), so we're getting somewhere. It seems to use only 4 of the loading bays, but if I use only 4 of them performance drops significantly. 6 seems to be ok, shifting the bottleneck to the junctions again.
Edit2: I've improved the station by creating separate paths for entering/leaving from/to N or S:
This further improves throughput to a train every 3 seconds (141 seconds for 50 trains), so that's pretty good I think. Total capacity would be 20*16=320k plates per minute, or the output of 1655 fully boosted smelters.
So, I think I found my station design. Any ideas for further improvement?
289 seconds for 50 trains with pre-stop, 238 seconds for 50 trains without the stop.
Edit: nm, there was a signalling issue. With 8 stops time is just under 200 seconds for 50 trains, adding a temporary stop does not seem to change much. The empty bay syndrom seems to be solved now there is enough room for a full train after the junction plus a signal just before
Test setup:
I added an extra track to the left to bypass waiting trains for trains exiting S (this solves the loop/deadlock problem), and made it a bit less compact to ensure a full train can wait between junction and first loading bay.
This is slighty better than the loop stations (a train every 4.5 seconds rather than every 6). Adding some extra signals gets it further down (to around 170-180), so we're getting somewhere. It seems to use only 4 of the loading bays, but if I use only 4 of them performance drops significantly. 6 seems to be ok, shifting the bottleneck to the junctions again.
Edit2: I've improved the station by creating separate paths for entering/leaving from/to N or S:
This further improves throughput to a train every 3 seconds (141 seconds for 50 trains), so that's pretty good I think. Total capacity would be 20*16=320k plates per minute, or the output of 1655 fully boosted smelters.
So, I think I found my station design. Any ideas for further improvement?
blueprint