My first drone-automated base

[Stickman]

Recovered the orginating thread title, because otherwise the whole discussion makes no sense anymore. See also comment at the last page -- ssilk

ignore my actual thread, pissing contest is below

---> Big update on Page 2 (direct link to post). <---



So, I've started and restarted and re-restarted until I'm blue in the face and love this game, but this is the first time I'm actually building beyond automating the science packs. I guess that makes me a Factorio newbie.

I have a solar powered factory going in 0.12. After using belts to finish off 99% of the tech tree, I decided to rip it all up and rebuild in a way that uses no belts or minimal belts and do so being as compact as is reasonable without sacrificing too much throughput or production. This base should be making very nearly all the base products I use or will use on a regular basis to expand the power grid as needed and to craft components used to build other bases and outposts.

My focus on this build was being as compact as possible in that, without looking TOO ugly, to have as many tiles as possible used for something functional.

How did I do? What would you do differently?

This is my base blueprint. Rotated in different ways (and furnaces swapped out for factories), I built most of the base around this common blueprint:



This is the entire base, not including outposts or solar grids. Reworking oil production is on the to-do list. I haven't seen many super-compact high-volume oil builds so if anyone has any ideas, I'm all ears. Err, eyes. Err...
Full size: http://puu.sh/jgjI5/de24592a25.png


Now, you can't have a drone-operated base without DRONES right? Here's the little factory group I set up to make drones:
Full Size: http://puu.sh/jgjPs/c57b4db677.png


I have iron and copper coming in via 3 trains. I have no main train station yet or anything complicated but I'm beginning to incorporate trains. You can also see part of the solar blueprint that I'm pretty sure everybody already knows about. The grid of smelters at the top of the frame is iron plates while the grid at the bottom of the frame is copper plates.
Full size: http://puu.sh/jgjuk/861f95b744.png


Here's steel, chests, roboports, turrets, ammo, and rails.
Full size: http://puu.sh/jgjwH/82145ad598.png


This is the center of my base, with 16 roboports for charging when I trigger all the factories working at once with robots flying everywhere. This base center manufactures a lot of the high-volume stuff I use so that robots making a lot of trips here with raw materials don't have to go very far no matter what direction they come from.
Full size: http://puu.sh/jgjzz/3f663543ed.png


This is green/red circuit manufacturing. I haven't taken the time to figure out how many reds I consume per minute on average so I'm not sure if I have the appropriate number of red circuit factories, but I'll be able to judge that the more I use the base. Same goes for green circuits. I just kinda eyeballed it. Definitely on the to-do list.
Full size: http://puu.sh/jgjCu/c5f1e13291.png




Prioritized to-do list:
1) re-locate solar power to be out of the way
2) evaluate green/red circuit production needs per minute, tailor assembler count to suit
3) Quadrant-based, raw item interlaced grid for entire factory, one entirely compact centralized square
4) Convert to smart inserters into provider chests
5) Stagger inserter quantities to avoid over production
6) re-work oil refining area to be more compact and allow for greater volume processing
7) build more walls, even though defenses are already fine

[DerivePi]

Smart Inserters would probably help.

[Stickman]

removed

[Zhab]

Your "factory modules" are quite compact and neat looking, but there is a lot of empty space in between sections for no apparent reason. If your goal is to have as compact a base as possible, you should probably work on that next. Considering that an electric furnace needs the same kind of setup/space as an assembler, there should be a way to place stuff in one huge perfectly aligned square of assemblers and furnace. With perhaps an occasional line for roboports or something.

There should only be 4 distinguishable sections to your base. Power, Assembler/Furnace, Oil Processing and train central. Ideally, those sections should be as close to each other as possible. That would definitively make for one of the most compact base ever built and I think it would also look pretty damn cool as well. But that is just my opinion, feel free to ignore it. I'm kinda obsess about making stuff compact after all.

But that being said, the closer stuff is from each other, the less distance your robots ever have to travel and the less often they have to recharge. Which should increase throughput of logistic network.

[Stickman]

There should only be 4 distinguishable sections to your base. Power, Assembler/Furnace, Oil Processing and train central. Ideally, those sections should be as close to each other as possible. That would definitively make for one of the most compact base ever built and I think it would also look pretty damn cool as well. But that is just my opinion, feel free to ignore it. I'm kinda obsess about making stuff compact after all.

Sweet! I like the 4-quadrant breakdown.

DEFINITELY going to add this to the to-do list. Thanks

[vampiricdust]

I personally put smart inserters on the requesters at set network wide limits. This way I can control how much of each I have. With .12 out you could use constant combinators to every assembler of an item in one place.

[Stickman]

removed

[Zhab]

For some reason, the entire concept of these combinators eludes me. I have not yet seen a good explanation of how to use them. Maybe I'm a dummy.

Combinator ? You don't need those. I'm guessing that you don't much about smart inserters either ? Just change the inserters you have that are pulling stuff out of assemblers to put them into provider chest with smart inserters.

smart inserter.jpg (29.06 KiB) Viewed 19157 times

What this says is only pull stuff out if there are less than 200 belts in the entire logistic network. You do the same for every assemblers making sure to select whatever items that particular assembler is making in the smart inserter settings.

You can choose any number that suit your fancy.

Combinators are for much more fancy stuff than this. People have bean doing lots of crazy things with just smart inserters and smart chests way before combinators became a thing.

[Stickman]

removed

[Zhab]

Yep, I know how smart inserters work. I used to use them a lot, but with limiting the number of available spaces in chests, I don't find a need to use inserters since they do basically the same thing and I'm not using them for complicated if/then triggers.

Although, as an example, I'd have to link all of the passive provider chests together with all of their inserters for the inserters to only extract from assemblers if network quantity is less than a certain number, right?

What if there's two assemblers each with their own smart inserter going into their own passive provider chest --and then there's a storage chest off somewhere else with several thousand more of the item? I'd have to link both inserters together with both passive provider chests with green wire (because green is better than red :p) and then they'd only work when the contents of those chests on the green wire network fit the equation, right? If I wanted that storage chest to also be taken into account, I'd have to link that storage chest with green wire, too?

You don't have to link anything. Roboport is already linking everything within range with the logistic network. By saying only have 450 electronic circuits in the system, that take into consideration everything connected to the logistic network (by roboports). This means all provider chests, all requester chest and even storage chest. With that kind of setup you should not even need storage chests because you never produce more than you need or at least not more than can fit in requester chests. For example, if you have 5 requester chests requesting 100 electronic circuits each, set your inserter limit to 450 or something and no circuits should ever make it to a storage chest. Unless you eventually manually dump 800 circuits in the system.

If for some strange reasons you have circuits in storage, they will act as buffers. Your provider chests should normally be empty. Because requesters can take 500 and the system is only allowed to create circuits if there are less than 450. So if requesters have room for 500 and only 450 are allowed to be produced, all circuits should be in requesters and absolutely none in providers. From this point, requester chests will be filled to 500 with what is available in storage chests until your stockpile runs out. Eventually your storage will be empty and as your factory burns trough electronic circuits, your circuit count will drop bellow 450. Which will allow smart inserters to pull new circuits out of assemblers and put them into providers. Which should be promptly emptied by robots.

That may sound complicate, but let me recap. The whole system will use whatever is in storage first before creating new items. Once storage is empty, new items are created at a rate that match the current demand.

Provider chests have higher priority than storage chest. So if you want your factory to drain your storage instead of slowly making it bigger and bigger and bigger over time, you need to use smart inserters to insure that your provider chests spend most of their time empty.

[johanwanderer]

A fully robot-transported factory. That's pretty cool. I usually have half of my factory running on belts, and the lazy half (for things I only need occasionally) run on robots.

But yes, use smart inserters to move products from assemblers into providers. Since everything is already connected as part of the logistic network, you don't have to wire anything. If I have multiple assemblers making the same thing, I also stagger the inserters settings, so less and less assemblers would be working once storage approaches the desired quantity.

For example, if you desire to have 2000 green circuits in the network, and you have 10 assemblers making them, I would have the first 2 stop at 1800, next 2 at 1850, next 2 at 1900, and the last 2 at 2K. That way, if the stored amount drops by a small amount, not all of them would spring into action.

[Zhab]

"Staggering" insertions is an interesting idea. Get production and power usage stable instead of having your factory "pulse" produce. Also makes it easier not to overproduce.

[Stickman]

removed

[Stickman]

removed

[DerivePi]

Can you create a "network within a network" using the colored wires? As in, force a smart inserter to only focus on the one chest it's linked to?

Yes - but that will require a red or green wire connection. Also note, the smart inserter conditions are "AND" conditions. As an example, if you have the Advanced Circuit connected smart inserter set to: robo net [ operate when robo net has less than 2,000 ADV CKTs ] AND red wire condition [ operate when connected chest has less than 200 ADV CKTs] - both conditions must be true for smart inserter to operate.

[Zhab]

That's a great explanation! Thanks! I thought I had to use colored wire everywhere for that kind of thing to work.

If all it takes is a roboport to link everything, are the green and red wires only really used when roboports aren't around?

Can you create a "network within a network" using the colored wires? As in, force a smart inserter to only focus on the one chest it's linked to?

The circuit network is many times more complicated to explain than the extremely simplistic logistic network (which are two different thing even if they both have network in the name). Especially in 0.12. Use to have only "item signal", but now we have number signal, letter signals and even color signals. Not mentioning the combibators and the shenanigans involving those.

The short answer is yes. If you link a smart inserter to a specific smart chest, then that smart inserter will only see that specific chest and will only react to the content of that chest. You would then have to place your condition in the circuit network instead of the logistic network.

That way you could have several separated CIRCUIT network side by side that are all interacting with different inserters/chests. You can connect the same thing to two different network by using green and red cables. And all those circuit networks could be contained within the same global logistic network (note that smart chests are not seen by roboports, but you can use robot chests in circuit network instead).

Finally, like DerivePi said, if you have a condition in both the circuit and logistic section both will have to be "true" for something to happen. In 0.11 you use to be able to have 2 circuit conditions by connecting both the red and the green cable to the same inserter. But in 0.12, you have to use combinators to handle multiple conditions setup and then output a simple true or false signal to the inserter to tell it to work or not work.

[johanwanderer]

Can you create a "network within a network" using the colored wires? As in, force a smart inserter to only focus on the one chest it's linked to?

As the posters above already said, you can definitely do that. However, another option is to separate your logistics into smaller, disjointed networks, and have pairs of requester/provider chests on the edge of the networks to pass things back and forth:

20150729-01-logistic-bridging.png

Notice that the left and right networks do not overlap. Also, for higher throughput (and waste), use a chest in between each pair of inserters.

The pictured networks pass a lot of things to one another. I use it to make sure I have essential items everywhere on the map, while making sure that my robots don't have to travel far. But you can adapt that paradigm to have, say a dedicated circuit-producing sub-network. Its edges would take in the essential stuff: iron, copper, coal, oil barrels, and provides red / green circuits and empty barrels.

[Stickman]

removed

[Tallinu]

... that take into consideration everything connected to the logistic network (by roboports). This means all provider chests, all requester chest and even storage chest. With that kind of setup you should not even need storage chests because you never produce more than you need or at least not more than can fit in requester chests.

Requester chests are destination-only, and bots will never remove items from them (unless you mark them for deconstruction). This means items in a requester chest are not available for request by the logistics system, and therefore do not count toward available items in logistics storage, whether for the purposes of smart inserter conditions or anything else.

This is easily verifiable: Place some items in a requester chest that are not present elsewhere in storage (maybe something like wood planks, or just set up an independent roboport with a couple chests), and mouseover a passive provider or storage chest. The requester's contents will not appear in the list. Requesting those items to yourself or another chest will not result in anything being delivered. And setting up a smart inserter in that network with a condition based on that item will result in it behaving as if there are zero items stored.

Also, you always want to have at least one storage chest, because otherwise deconstructed items or aborted requests (like when you run out of range) will have nowhere to be stored and your bots will simply hover in place carrying the items forever and running out of charge (at least, until something eventually requests what they're holding).

... Your provider chests should normally be empty. Because requesters can take 500 and the system is only allowed to create circuits if there are less than 450. So if requesters have room for 500 and only 450 are allowed to be produced, all circuits should be in requesters and absolutely none in providers. From this point, requester chests will be filled to 500 with what is available in storage chests until your stockpile runs out. Eventually your storage will be empty and as your factory burns trough electronic circuits, your circuit count will drop bellow 450. Which will allow smart inserters to pull new circuits out of assemblers and put them into providers. Which should be promptly emptied by robots.

This makes me wonder if you've been using only active providers, without realizing the difference between active and passive. It's not an accurate representation of how the system typically works. Maybe part of that is due to the mistaken understanding of requesters.

Since the contents of requester chests don't show up as available stored items, a smart inserter set to work until there are 500 circuits will drop items into its provider chest until all requests have been fulfilled and there are an additional 500 circuits stored, counting all storage chests and passive provider chests within the network, not just the one the inserter is filling. If all the circuits keep getting pulled out, that just means your system is consuming them faster than you are producing them -- or your requester chests are set to request more items than necessary. Since items in one requester are not available for use anywhere else, the requests should be just large enough that the amount stored will not be completely used up before bots can arrive with another delivery of materials. The best quantity to request is affected by a number of factors, including the typical rate of consumption of the items, the distance between the usual source (provider or storage) and the requester, your logistics bot speed and carrying capacity, and the number of available bots (ones that aren't busy doing other work). It often takes some trial and error, and typically, a slightly higher than necessary request amount won't do any harm -- but requesting, say, 1,000 or more of something is (in typical situations) massive overkill.

The whole system will use whatever is in storage first before creating new items. Once storage is empty, new items are created at a rate that match the current demand.

This manages to be correct even though the explanation is different - as storage empties out, items will be produced and typically wait in passive providers, and once storage of that item is empty, requests will pull the items from the providers instead of storage. The smart inserters will then refill the providers from the assemblers (up to the set network-wide limit) as quickly as the assemblers can work, so even if the providers get emptied out by a sudden large request, they shouldn't stay empty for long unless production simply can't keep up.

Provider chests have higher priority than storage chest. So if you want your factory to drain your storage instead of slowly making it bigger and bigger and bigger over time, you need to use smart inserters to insure that your provider chests spend most of their time empty.

Active providers do have a higher priority than storage, and in fact, even if nothing is requesting their contents, they will be actively emptied into any available storage space. This gives them a high potential for overproduction (and potentially filling up all your storage) if you fail to assign correct logistics conditions on the smart inserters putting items into them. Setting a hard limit by redding out all but one slot instead of using smart inserter conditions simply doesn't work, since that slot will still be continually emptied no matter how many items end up in storage.

Passive providers, on the other hand, have a lower priority than storage chests. As mentioned above, requested items will be pulled from storage first (if any are present in storage) before being pulled from passive providers, to avoid having items build up in storage and never get used. Passive providers are what most people typically use for the output of assembling machines in the main parts of a factory. And in my opinion, they are far better -- unless a provider is located at a great distance from the place(s) its items are generally consumed or requested from, while your storage chests are very close to that area. Having the items actively transferred to storage would then mean that requests, when they come, are filled much faster.

Keep in mind that using active providers instead of passive causes each item to be moved twice by your logistics bots, once when produced and again when requested. Heavy use of them can potentially double the amount of work your bots are required to do, increasing power consumption and reducing number of available bots, as well as forcing more storage space to be consumed by items that could otherwise simply wait in the providers at their factories.

[Stickman]

ignore my actual thread, pissing contest continues

I've had some time to rework my little base and take some screenies.

I still haven't figured out the green-to-red circuit ratio for my needs, but that won't be hard to do once I start ramping up production when the base is finished. Then it's just a matter of adjusting which factories are doing reds vs greens in the existing layout.

I've upgraded all my passive provider chests to be fed by smart inserters. I've tailored many of the smart inserters (for items that have many assemblers making said item) to have staggering working rules so I don't mass produce a ton of product if the inventory drops below the threshold by, say, one single unit. I plan to continue this staggered output method more aggressively as the base and outposts grow, but for now when I only have a few factories making things it won't break the bank if, say, all the green factories kick in to run a few cycles and then shut off if inventory dips 1 below threshold. That'll certainly come in handy later, though. Love that suggestion.

I haven't done much work to the oil refinery system yet. That's a big project but it's the next big project on the list, before really reworking trains. I want to bring oil barrels in on trains so I'm trying to think ahead in how to incorporate specific train cars for specific items to offload properly in relation to where my oil refinery project ends up being located.

New Base Layout with mentioned upgrades. You can see the budding train station down below. I want to have two incoming platforms for iron and two for copper. I'm still not entirely sure how I want the train station to work, so for right now I have one train running each track.
Full Size: http://puu.sh/jjGGn/73460d6d72.png


A little better lighting for the screenshot.
Full Size: http://puu.sh/jjGti/a3823d2230.png


I put half the iron plates on one side, half on the other. I did the same half/half split with copper plates and steel production. I filled the "bottom center" areas with some common goods like gear wheels and copper wire, though the group of 4 items just to the left of center I didn't put too much thought into. I can probably trade places with some other kind of product that would be better suited to such a close proximity to the smelteries. I'll have to consider fixing that later.
Full Size: http://puu.sh/jjGyR/b008ebf39f.png


This is my green/red circuits with plastics and batteries squeezed into the side, along with belt production. This is the right-side of my main base.
Full Size: http://puu.sh/jjGBr/dd19723a4e.png


This is the rest of the bits and bobs for most common items I'll be using to mass-construct outposts, defenses, etc. I think I'm going to swap out one "set" of assemblers in this shot that are making solar panels to make electric mining drills instead. I forgot to do that earlier. This is the left-side of my main base.
Full Size: http://puu.sh/jjGDh/8c17188e96.png