Factorio Forums

This is more of a discussion at this point.

1. I have moved backed to belts for outposts. I know for larger designs this isn't the best for UPS but it has an advantage now with BP books that I can mine resources w/ much less power required. I used red belts on this play through. With full productivity on the gears..they aren't that much more expensive that yellow. If I use 4 car trains this means at most I can pull out 106.664 ore per second. With 60 updates per second and a free running counter, it is possible to turn a belt on off with just better than 2% steps (as low as 1.77 ore/sec from an outpost). This got me thinking...I have a granular control variable I didn't have with bot outposts.

2. Most systems when they go idle land up basically compressing all the way back. In the case of very large train system, this means, unless your stacker is stupidly huge, you land up having a backup onto the main line (e.g. what happened in the 0.15 sim series). The second negative of this is that it takes longer to find out your bottlenecks since you need to wait for the backlog to clear out. We can also better utilize the stackers that occur at remote outposts.

3. If we eliminate the buffer (minimize) chests at outposts and add as many trains as it takes to allow those belts to never stop, then we have a delay we can work with (lets say it is 1 minute). This also allows us to aggregate our throughput. If had 20 outposts..I could in theory do ~2133 ore/second. I do think this does bring up an issue of you need to know your smelter. If you can supply ore faster than you can consume it then you need to limit the absolute throughput. So if I can produce 16 blue belts of iron plate on a PM3 system then I need to limit ore to 533.333 ore/second or and approximate duty cycle of 25%.

4. If we do this we can no construct a "model" of our system and now apply controls to it. This is a rough sketch I did in a few minutes and I think it is possible to simulate this and come up with a possible PI controller in factorio which is "tunable" to your setup (e.g. how many outposts, what kind of belts , smelter size). This same logic can also be applied to other areas of high traffic or resource usage (like green circuits and steel). The integrator probably also needs limiting but that is another issue that isn't too hard to deal with.

Thoughts or should i just continue to make mega stackers for my 20+ outposts and 50+ trains just for iron?

http://imgur.com/a/eXKYO

Stackers are the easy solution. There are other options, involving not sending out trains when you don't need resources.

sparr wrote:Stackers are the easy solution. There are other options, involving not sending out trains when you don't need resources.

But if forces you to put the stacker at the end where it can back out onto your main rail line if it isn't big enough. That is the problem i'm trying to avoid Having a stacker that hold 50+ 2-4-2 trains seems a little along with the required stackers at the outposts; that seems like using a hammer instead of a scalpel. If you would like to share more robust (complete) ideas then please do. i would love to hear something other than the obvious.

Well, if watch your resources either by counting them on the belt (after the smelter) or in the logistics network and connect your smelter with all outposts, you can hold the trains at their outposts until the smelter begins to run low.

Assuming that each train only serves one specific outpost, this reduces the size of your smelting-stacker and keeps the ones on the outposts pretty small, too (Number of trains at outpost - 1).

Another approach would be to create a yard near the smelter that holds every ore train after unload until resources run low and it needs to go back to its outpost. This would keep the amount of circuit wires lower but increase the calculation-complexity since trains would basically need twice the time. Of course, this would also just be one giant stacker but could be decoupled from the smelter or outpost itself.