Automated Factorio Thread [TODO: add automation]

User avatar
Retsam

Automated Factorio Thread [TODO: add automation]

Postby Retsam » Mon May 15, 2017 4:20 pm

I know we don't have extensive Factorio discussion going on in the "What we're playing" thread, but I know there's a number of us who play it, and I figured it might merit it's own thread, anyway, if for no other reason, so that when someone posts a link to some useful resource I'll be able to find it later.

To that end, I'll try to keep this first post up-to-date with whatever useful links people post in this thread.

(Also, if someone has a better idea for the thread title...)

Resources:
Factorio Train Automation: From Novice to Journeyman (Basic guide to trains)
Stations, Junctions, and all things Deadlock (Advanced guide on preventing deadlo)
User avatar
Retsam

Re: Automated Factorio Thread [TODO: add automation]

Postby Retsam » Mon May 15, 2017 7:49 pm

I finally managed to finish a game of Factorio, now that I've launched a rocket from my .15 world.

This is my fourth factory; though the middle two were pretty short-lived (the second was abandoned when I decided to play with trains, and the third was made obsolete by the .15 update). My first factory died in the "3 x 100 rocket parts" slog, since things were too spaghettified and there were too many manual steps to my factory.

Anyways, .15 is really nice. Smoothing out the science curve was a great change, fluid cars are a godsend (as someone who used to have an awkward pumpjack > barrel > train > un-barrel > refinery system setup on a previous factory), and just a number of little tweaks and improvements here and there.

Nuclear Power

I was in agreement with Shamus on nuclear power; by the time I got it going I could have already laid down thousands and thousands of solar panels, making nuclear power a pretty unnecessary toy compared to solar/steam.

The main advantage I found using nuclear was that I had enough of a power surplus that Electric Furnaces felt a lot more viable, which simplified things quite a bit in a couple cases. For example, I made a off-site electronics factory by throwing electric drills furnaces on top of a conjoined copper/iron patch, constructing the circuits, then shipping them in by train. Otherwise I would have had to ship in coal, or else ship the raw resources off to a secondary site to do the smelting before making the chips. (Which adds a bunch of places things can go wrong)

I said "was in agreement", though, because what I did (and I assume what Shamus did) was wait until I had stockpiled 40 enriched uranium and had Kovarex enrichment researched, then setup an enrichment pipeline, and then started using nuclear power. If you do this, then, yes, you're going to be very late into the game by the time you get nuclear power running (as you need 1000 of each science pack for this research, plus it takes a long time to get 40 enriched uranium)

On the other hand, I found a reddit post today that suggests that you shouldn't actually wait until you have Kovarex enrichment going; you can start using nuclear power almost immediately after researching it, and it only takes up to blue science (which is much easier in .15) to research.

Apparently, even just a single centrifuge (albeit one running at full load constantly) is enough, just barely, to keep a single reactor running. You'll want a small stockpile of enriched uranium to assure you don't get too unlucky and run out of fuel cells, but nowhere near 40 (2-3 seems to be the consensus in the reddit thread).

It'll take a lot longer to reach the Kovarex threshold if you're using uranium as you get it, but the tradeoff seems of nuclear power early (while it's still useful) seems more than worth it. I'll certainly be testing this approach in my next game.
User avatar
dudecon
Location: Camarillo, CA. Paul Spooner IRL & blog comments
Contact:

Re: Automated Factorio Thread [TODO: add automation]

Postby dudecon » Mon May 15, 2017 9:40 pm

I really want to share all my tips and stuff... but I've never played. I'm sorely tempted to buy a copy, just so I can contribute, but I'm already pegged for time and energy.
Also, my day job involves a lot of routing and sizing, and this kind of sounds like more work.
User avatar
Trix2000
Location: California

Re: Automated Factorio Thread [TODO: add automation]

Postby Trix2000 » Tue May 16, 2017 9:18 pm

Retsam wrote:This is my fourth factory; though the middle two were pretty short-lived (the second was abandoned when I decided to play with trains, and the third was made obsolete by the .15 update). My first factory died in the "3 x 100 rocket parts" slog, since things were too spaghettified and there were too many manual steps to my factory.
I think I'm on my... sixth now? Might be seven, and I'm not including the one I worked on with a friend which got most of the way to the rocket but didn't finish.

For me, the rocket is something I usually just do when I run out of other things I want to improve... though with the change to science I've actually decided to use my mess of resources to automate launches (because infinite research is kinda neat). I'm at the point where I think really large-scale fairly early, to the point where I actually had some significant biter attacks in my 0.15 playthrough due to scaling up production so much and so quickly (without efficiency module production to temper my pollution).

I took that factory the farthest I ever have, with dedicated production outposts for iron plates, copper plates, steel, and green circuits. It's enough production that I can actually automate rocket parts/launches faster than I can use the space science.

Anyways, .15 is really nice. Smoothing out the science curve was a great change, fluid cars are a godsend (as someone who used to have an awkward pumpjack > barrel > train > un-barrel > refinery system setup on a previous factory), and just a number of little tweaks and improvements here and there.

I feel like the science got both less and more complicated, both in good ways. Blue science is now a reasonable step up instead of being a significant leap, and there's actually some interesting management in the higher level science packs instead of just killing a billion biter nests and calling it a day. I actually found myself prioritizing the high-end packs based on what I could produce and what technologies I need, which is a much-needed thing I think.

I do like the fluid wagons, but I actually think they're about on-par with barrelling which is good... they each have advantages and disadvantages. Fluid wagon is easier setup and management (just top up the tank), but carries less per wagon. For transporting crude oil it's particularly nice since I don't need to bother with barrel management early.

Nuclear Power

I was in agreement with Shamus on nuclear power; by the time I got it going I could have already laid down thousands and thousands of solar panels, making nuclear power a pretty unnecessary toy compared to solar/steam.

It takes a LOT of time and material to create those panels and accumulators, though, and a TON of space. Nuclear's biggest advantage is, once you get it rolling, it's very very easy to expand it to provide extraordinary amounts of power. A 4-reactor setup provides 480MW on its own, which way more than enough power for the rocket (not for me, though, I'm running something like 12-16 reactors now). It's incredibly efficient too, and no pollution!

Steam is also easy to expand and cheap, but provides much lower amounts of power. I basically relied on it alone until I could get sustainable nuclear (with Kovarex), because with the boiler changes it's really simple to add more.

Solar is great because it's fire-and forget, though expansion can be a pain without robots and even then can take time to set up very large amounts of power for lategame. Not to mention the amount of space required to generate hundreds of MJs of power.

The main advantage I found using nuclear was that I had enough of a power surplus that Electric Furnaces felt a lot more viable, which simplified things quite a bit in a couple cases. For example, I made a off-site electronics factory by throwing electric drills furnaces on top of a conjoined copper/iron patch, constructing the circuits, then shipping them in by train. Otherwise I would have had to ship in coal, or else ship the raw resources off to a secondary site to do the smelting before making the chips. (Which adds a bunch of places things can go wrong)

I pretty much go for electric furnaces as soon as possible, because dealing with coal limits my setups somewhat. It's easier to manage their power use by throwing a couple Efficiency 1 modules in them (though the miners should get those first).

I said "was in agreement", though, because what I did (and I assume what Shamus did) was wait until I had stockpiled 40 enriched uranium and had Kovarex enrichment researched, then setup an enrichment pipeline, and then started using nuclear power. If you do this, then, yes, you're going to be very late into the game by the time you get nuclear power running (as you need 1000 of each science pack for this research, plus it takes a long time to get 40 enriched uranium)
I think I might have a different definition of 'lategame' than you. :P

It's incredibly worth it to get that running, though, because the amount of fuel you can get from even a small patch of uranium can last for hours. I stockpiled so much that I could probably run my current reactors for literal weeks, and I still have a 5-6mill uranium patch waiting to be processed which is incredibly overkill.

On the other hand, I found a reddit post today that suggests that you shouldn't actually wait until you have Kovarex enrichment going; you can start using nuclear power almost immediately after researching it, and it only takes up to blue science (which is much easier in .15) to research.

Apparently, even just a single centrifuge (albeit one running at full load constantly) is enough, just barely, to keep a single reactor running. You'll want a small stockpile of enriched uranium to assure you don't get too unlucky and run out of fuel cells, but nowhere near 40 (2-3 seems to be the consensus in the reddit thread).

It'll take a lot longer to reach the Kovarex threshold if you're using uranium as you get it, but the tradeoff seems of nuclear power early (while it's still useful) seems more than worth it. I'll certainly be testing this approach in my next game.

I might consider that, but the one thing that concerns me with that is it makes getting the initial 40 take that much longer... and you DO want to get Kovarex running eventually. I was able to cover the difference with steam (solar too if I was using it, but I was going for the achievement) so I don't know if it's too critical, but then even just the base 40MW is a nice boost early on.

Ultimately I'm incredibly pleased by nuclear implementation though. It's an interesting challenge to set up and get rolling properly (and making sure it can't lock up or lose your 40), but once you do its very easy to throw down incredibly excessive amounts of power. Given that I'm using something like 500-700MW on robots alone right now... it helps a lot.


Here's a picture I had of my 0.15 factory, though it's a little outdated (I've finished wallling off the West bit you can see turrets in).

Also, nukes are fun.
User avatar
AileTheAlien
Location: SK, SK

Re: Automated Factorio Thread [TODO: add automation]

Postby AileTheAlien » Sun May 21, 2017 11:12 pm

Retsam wrote:fluid cars are a godsend


One other thing that fluid cars make viable, and actually a lot less headache than the alternative, is long-distance electricity. I'm playing with vanilla, and rail-world settings that are even more rail-worl-y than normal. Even before I started using trains heavily (from train-world settings in 0.15 and/or the Resource Spawner Overhaul I used before 0.15, which is even more train-world-y), I found it tiring to place so many "large" electric poles. They only reach 30 tiles, so my base would be about 10 poles wide, and the extended laser-turret perimeter would extend that diameter to something like 20 wide. Even though I only needed to cover the diameters and circumferences with large poles, my base needed several hundred of them. Fluid trains with one or two steam wagons makes that uneccesary. Just one tanker is enough for an efficiency-moduled mining operation, and later once I get to nuclear, the hotter steam will pack even more power. :)


Retsam wrote:The main advantage I found using nuclear was that I had enough of a power surplus that Electric Furnaces felt a lot more viable

Trix2000 wrote:I pretty much go for electric furnaces as soon as possible [...] throwing a couple Efficiency 1 modules in them (though the miners should get those first).


I normally go electric smelters, because they can be efficiency-moduled, which means they burn less coal (at the power station) than steel smelters burning coal themselves. With my current train-world settings (turned down even more to..."low" or maybe "very low" resources? lowest non-zero settings for aliens), coal is even more scarce than normal, so efficiency modules/electric is even more of a necessity. :)


Retsam wrote:suggests that you shouldn't actually wait until you have Kovarex enrichment going; you can start using nuclear power almost immediately after researching it, and it only takes up to blue science (which is much easier in .15) [...] Apparently, even just a single centrifuge (albeit one running at full load constantly) is enough, just barely, to keep a single reactor running.


This sounds pretty good, actually. I'm almost at the stage of fully train-automating my research and circuit production, so maybe next I can get some basic nuclear going. I just cleared out some aliens from a 16-blob oil patch, which will bring up my total number of pumpjacks from 3 to 19. I'm currently just barely getting enough fuel from advanced oil cracking, to run my efficiency-moduled base / boilers, so this should tide me over until I get nuclear going. (My 3-blob patch is already down to the 20% low-production time.) I think the only thing saving my current game, is that oil is infinite. If it was limited (which I actually want put into vanilla), I guess I'd have run out of resources long ago, trying to make enough solar panels. :)
User avatar
Retsam

Re: Automated Factorio Thread [TODO: add automation]

Postby Retsam » Mon May 22, 2017 1:44 pm

The "super trainworld" settings sound interesting. I do imagine it would change the game quite a bit to have to deal with proper scarcity. I'm planning my next world (which, for sake of my free time, is not in the immediate future) to increase the alien difficulty (perhaps one of the two combat-oriented presets), and probably shouldn't combine that with resource scarcity, unless I really want to have a bad time.

I need to get a better grasp on the steam mechanics; I know that people use "steam batteries" and such, but I usually struggle to just figure out exactly what's going on with my power plants, never mind actually trying something complicated.
User avatar
AileTheAlien
Location: SK, SK

Re: Automated Factorio Thread [TODO: add automation]

Postby AileTheAlien » Mon May 22, 2017 4:02 pm

Retsam wrote:I need to get a better grasp on the steam mechanics; I know that people use "steam batteries" and such, but I usually struggle to just figure out exactly what's going on with my power plants, never mind actually trying something complicated.

It's not too complicated, if you just break down power production into small pieces. After that, it's like playing with LEGO - just mix and match the parts, until you have something that works.

Before fluid-handling / trains:
  • Perfect ratio for water pumps -> anything else: Ignore this. Pumps are dirt-cheap, so getting optimized water is for weirdos to waste time with. :)
  • Perfect ratio for boilers -> steam engine: 1:2 in version 0.15 of the game. This uses up all the steam/heat, without extra or shortage. (This number is basic math from the numbers listed in-game for steam engines and boilers.)
After fluid handling and trains, you get this stuff:
  • Boiler -> pump -> steam tank: This is a huge buffer of energy, although not as large as a chest full of wood / coal / etc. Less hassle to set up / build, though, since you need less pumps than you would need inserters. :)
  • Steam tank -> pump -> other steam tank: Useful for separating your large steam-buffer(s) from your train-loading/-unloading tanks. Big buffers can be made as large as you want, but for super-fast train-loading/-unloading, you want 3 pumps per wagon (the max), and one tank per pump.
  • Steam tank -> pump -> train fluid wagon: If you put the tanks right beside the train-pumps, you'll get a full fluid wagon in a couple seconds, instead of a minute. Pipe flow is slow for long distances. Again, 3 tanks / pumps per wagon is the maximum speed.
  • Train fluid wagon -> pump -> tank: Steam to remote areas. Again, put the tanks right beside the train-unloading pumps, so you can empty the train in seconds.
  • Steam tank -> pump -> steam engines: Turn the steam from your remote areas into electricity. Needs a single solar panel to kick-start the pumps, before the steam engines are loaded with hot steam.:)

I figured this stuff out myself, and it's not necessary to memorize any particular layout, unless you like symmetry. The main things to remember with fluids are:
  • Pipes flow slowly over distance, so put a pump every once in a while, or keep distances short.
  • Tanks and pipes flow slowly based on how full they are, so pumps are needed to suck or push fluid to where you need full-speed.
User avatar
Trix2000
Location: California

Re: Automated Factorio Thread [TODO: add automation]

Postby Trix2000 » Mon May 22, 2017 5:23 pm

AileTheAlien wrote:[*] Perfect ratio for water pumps -> anything else: Ignore this. Pumps are dirt-cheap, so getting optimized water is for weirdos to waste time with. :)
Hey! I resemble that remark!

Though I think it's good to know the number (1:20 btw), because it means you know exactly how much you can get out of a given patch of water - since space around the water's edge may be limited depending on circumstances. This, the most you can have in an optimal basic steam engine setup is 1 pump : 20 boilers : 40 engines. I usually like to split it into two opposing rows of 10 boilers, with two engines on each.

  • Pipes flow slowly over distance, so put a pump every once in a while, or keep distances short.
Pumps will maximize flow rate now, so if you need to pipe something longer distance you can use them to boost the flow rate. Additionally, underground pipes only count as ONE piece of pipe for the purposes of distance, so it's best to rely on them as much as you can - you can go quite a long way before needing pumps with them. In fact, if you're willing enough to lay the pipes, you don't technically need trains to move fluid since you can maintain flow indefinitely (though not recommended much).

Though I personally don't feel sending steam by train is any more advantageous than running power lines, since running power lines is easy (click and drag, or stick it in your rail blueprints) and probably isn't much worse than setting up the pumping stations on either end. Not to say it's something that shouldn't be done (because it's fun) but I don't feel it's a much easier or better solution. More interesting, to be sure.

  • Tanks and pipes flow slowly based on how full they are, so pumps are needed to suck or push fluid to where you need full-speed.

Depends - much of the time, a pump is not needed to keep things moving, as fluid flows fairly naturally in smaller setups. It's only really necessary to include pumps when distances start to get high or you really need high throughput (though it often doesn't hurt to throw extra pumps in anyways).



Retsam wrote:I need to get a better grasp on the steam mechanics; I know that people use "steam batteries" and such, but I usually struggle to just figure out exactly what's going on with my power plants, never mind actually trying something complicated.

Steam is just heated water, so it functions in pretty much the same way (including the amounts). An offshore pump provides 1200 water/s, and engines consume 60/s. I'm not certain on the kind of throughput boilers and heat exchangers can manage, but I don't think it comes up much (if at all).

One thing that may seem complex is how steam should be handled for nuclear, since the numbers are less obvious. One thing that can trip things up is making sure you have enough water for your setup, since turbines consume the same amount of water/steam as steam engines (60/s) and larger nuclear setups will need more than one water input (1 per 40 turbines I believe). At that point, you can calculate the number of heat exchangers and turbines fairly easily if you know the total MW output of your reactor(s):

- Each reactor gives 40 MW by itself. For each active reactor next to it, you add another 40 MW. So two adjacent reactors would each have 80 MW (160 total), and four in a group would be 160 MW each (480 total).
- Each heat exchanger consumes 10 MW of heat. So take the total from your reactor, cut the 0, and you have the number of heat exchangers.
- Turbines are a little trickier - each will use ~5.8 MW of power. Divide the total reactor output by this to get the number of turbines you need - round up if necessary so you don't waste that last little bit.

Once you have those numbers, you just need to set these things up accordingly... although it should be noted that heat pipes start losing effectiveness at longer distances (it might be good to divide up your heat exchangers into separate columns) and there might be considerations to be made for flow rate... but this is really only of concern for larger setups.

As for storing steam from this... I personally stopped bothering since I have an excess of nuclear fuel and use most of my capacity anyways, but it may be as simple as throwing some tanks in between the exchangers and turbines. Wire one or more of the tanks to the inserters that insert fuel to the reactor, and have them turn off when the tank(s) are almost full. Since reactors don't cool off naturally, they will remain at 500 degrees even without fuel and not require a long restart time.

It might also be noted that I use no regular pumps in my nuclear setups. They're not really needed since the distances aren't all that long.
User avatar
Retsam

Re: Automated Factorio Thread [TODO: add automation]

Postby Retsam » Mon May 22, 2017 5:35 pm

Trix2000 wrote:
Retsam wrote:I need to get a better grasp on the steam mechanics; I know that people use "steam batteries" and such, but I usually struggle to just figure out exactly what's going on with my power plants, never mind actually trying something complicated.

Steam is just heated water, so it functions in pretty much the same way (including the amounts). An offshore pump provides 1200 water/s, and engines consume 60/s. I'm not certain on the kind of throughput boilers and heat exchangers can manage, but I don't think it comes up much (if at all).


Yeah, I guess I should have clarified, it's not that the .15 update introducing "steam" that confused me, it's that I've never quite understood the pipe or the waterheating mechanics in the first place. I'd find my plant operating with less than full potential and basically just threw stuff at the problem (more pumps, more pipes, more boilers, whatever) until it went back up. It got better once I found the 14:10 ratio, but I still didn't really understand it.

But, yeah that nuclear setup stuff looks really helpful. I never needed much power, so I never really got beyond the "throwing stuff at a wall and seeing what sticks" stage in my last game.
User avatar
Trix2000
Location: California

Re: Automated Factorio Thread [TODO: add automation]

Postby Trix2000 » Mon May 22, 2017 6:35 pm

Retsam wrote:Yeah, I guess I should have clarified, it's not that the .15 update introducing "steam" that confused me, it's that I've never quite understood the pipe or the waterheating mechanics in the first place. I'd find my plant operating with less than full potential and basically just threw stuff at the problem (more pumps, more pipes, more boilers, whatever) until it went back up. It got better once I found the 14:10 ratio, but I still didn't really understand it.
Yeah, the fluid mechanics aren't the easiest to figure out at a glance. I had a reference for them before 0.15, but the new patch made some tweaks to how fluids handled (mostly affecting pumps) that influenced things. It's not completely different in practice, but it does mean not having to do certain things (like no longer needing several parallel pumps to get higher flow rates).

Thankfully, if you keep to smaller setups, much of the fluid mechanics stuff won't affect things and the only important considerations are input rate (how much water you have) and output rate (how much water/steam your engines/turbines use). Pumps and considering flow shouldn't matter much for a basic one-offshore-pump system since a single pipe can hold a LOT of water and the throughput should be fairly constant.
User avatar
AileTheAlien
Location: SK, SK

Re: Automated Factorio Thread [TODO: add automation]

Postby AileTheAlien » Sun Jun 11, 2017 7:27 pm

Having gotten relatively far in my current train-world game[1], I have to say that I like trains a lot, although I think I've finished the game, despite not being at rocket-launch yet. Once you get your trains set up, and know how to signal and schedule them, you've essentially got a high-throughput, easily-expandable logistics network. It just uses more space, less energy, and doesn't require the research of robots. It doesn't replace robots, since those just remove the need for spaghetti conveyor belts for low-use items. It's sort of the same problem for the whole game - it's pretty easy to figure out the tricks for each piece of the game, and then you're done learning. It's still fun to mess around with the game at that point, but it stops being a puzzle game, and just becomes a model train set. (Pretty literally, for me.) :)

[1] And having fought with the BS size imbalance of alien bases. i.e. The setting the frequency low for aliens also increases the size, because these settings just mess with the random noise function - lower frequency stretches out the function, so the total amount of area in the function that's above the threshold for "aliens here" becomes much larger. :C
User avatar
Retsam

Re: Automated Factorio Thread [TODO: add automation]

Postby Retsam » Mon Jun 12, 2017 1:34 am

You missed one point, compared to logistics robots, trains use more space, less energy, and are infinitely more deadly. I was killed far more often by my own trains than anything else, in my last world.

I guess the other part of my last world that I didn't mention was that I avoided logistics bots. And it really didn't end up being that painful, just leaving myself a bit more space to work with. They're nice to have, obviously, but they aren't quite as essential as I might have guessed.
User avatar
Trix2000
Location: California

Re: Automated Factorio Thread [TODO: add automation]

Postby Trix2000 » Mon Jun 12, 2017 7:17 pm

The main challenge once you get to the level of using trains is more about scaling everything up, which introduces some new problems you may not have run into before. For instance, you'll find that inserter speeds and capacity are much more relevant, module use and production becomes more important, single belts stop being enough to supply certain things, and a lot of the original smaller designs you made just won't scale up well and may have to be rebuilt. Power may also be of concern, unless you have nuclear or a ton of solar laid out (especially if you rely on robots a lot, because they can eat tons of power).

It's what makes the rocket actually a fair decent challenge - it requires LOT of materials to make, so if you want to get one launched in any reasonable amount of time you have to scale things up well. Circuit production in particular is something you'll want a lot of, and that means making sure your green/red circuit production is large and well-fed.

And of course, there's always science to work on. Scaling up production of that can speed up your remaining technologies, as well as making the infinite techs easier to get (once you can automate rocket launches, which is another fun challenge).

Return to “Videogames”

Who is online

Users browsing this forum: Majestic-12 [Bot], mwchase and 2 guests