Do you do integrated production? And how much of it?

[Laie]

I'm currently still doing everything in my "initial" factory; the plan is that I move out all the production for science packs to dedicated plants where I can set things up such that everything will be produced in matching amounts; in due time, the starter factory is supposed to serve as mall and nothing else. It may still choke when I pick up a large delivery of blue belts, but this will no longer affect the production of everything else.

To that end, I've started drawing up plans for a major circuit plant. Lots of copper, iron and plastic go in, while blue circuits and Rocket Control Units go out. Oh, and a quite a few red circuits, too. Tweaking the outputs to matching proportions in non-trivial but possible. However... once built, it's essentially immutable. I cannot go back and change anything as that would change everything, which probably means that the only change I may ever make to that site is decommissioning it completely. If it every comes to that. Which it really shouldn't, seeing that I planned it so well -- but in my experience with playing Factorio, I've noticed a certain tendency for even the most cunning schemes to become obsolete after a while.

I'm starting to wonder whether a large integrated factory is even a good strategy, or if I should rather go for a more modular approach?

[astroshak]

Honestly?

I try for the more modular approach. Though, I should mention I’ve yet to actually finish a megabase beyond just a belt-driven 1k SPM base.

How much science/second (or per minute) are you looking for? I have been trying to start with the Rocket Fuel. Make a fair amount extra, as this is also used to fuel trains. I then make Green Circuits, Red Circuits, Blue Circuits areas. A module making area to drop off modules (consuming those GC, RC, BC produced for science) for the rest of the areas then follows, though I don’t drop off more than a stack each at any one area. Then the various types of Science, and labs to drink those beakers.

I usually get bored with it around Yellow or Purple science, that’s why I’ve yet to actually finish a larger megabase.

[Trific]

The only integrated production in my current build is firearm magazines with piercing rounds magazines because they are a perfect ratio (1 FM to 3 PRM), and firearm magazines aren't used anywhere else. Everything else is modular. The lab module consumes 896 SPM on follower robot count (half that with any other infinite tech). I have a bot mall to produce building materials that aren't part of the science pack line, but all of the basic materials for that comes from the wider base, so it's very easy to just slap down extra modules (beyond computed requirements for the rated SPM) to meet that extra demand.

I would recommend your first megabase be more modular, simply because you can do a lot of tinkering without having to rebuild the whole thing. I substituted infinity chest production for the modules I hadn't yet designed during the design and testing phase and worked my way up until I had blueprinted modules for everything required for the full base.

That said, I intend at some point to build a fully integrated base, which I expect will be a fun challenge for all the reasons you cite.

[eradicator]

Integration takes longer to plan, but other than that I think the main difference is what kind of blueprint you end up copy/pasting in the end. Because you will always be c/p'ing something. It also depends on the level of integration.

Here's one of my personal experiments with ore->science direct production. Pipes/gears aren't direct here because the ratio is order of magnitudes smaller than everything else. Also this isn't "perfect ratio" because that contradicts full integration :p. Haven't gotten to rockets yet cos everything took ages to plan.

screenshot_065h_46m_25s.jpg (337.24 KiB) Viewed 3140 times

[Laie]

How much science/second (or per minute) are you looking for? I have been trying to start with the Rocket Fuel. Make a fair amount extra, as this is also used to fuel trains. I then make Green Circuits, Red Circuits, Blue Circuits areas.

Hmmm. I notice that, perhaps, I should have spelled out what I mean when I say "integrated". Not sure if this conflicts with established lingo, but when I talk of a (sub-)factory I mean a local concentration of machinery that uses belts for transport, while individual factories are interconnected only by rail. Everything together is "the Factory" with definite article and capital F.

An integrated factory, then, would be one that produces several outputs from a shared pool of intermediates: in my case, heaps of green circuits going towards red and blue circuits as well as modules (and thus ultimately to rocket control units). To repeat, I guess the key point is not the shared pool but that there are several outputs which need to be balanced against each other.

I guess that ultimatley it doesn't matter very much, except...

I think the main difference is what kind of blueprint you end up copy/pasting in the end. Because you will always be c/p'ing something.

Well, if you put it thus... in the longer run it will probably be better to have factories dedicated to exporting either this or that, rather than making a bit of either. If only to reduce the number of train stations where any given item may be picked up.

[astroshak]

It seems to me the question you were really asking is ... when making (for example) Rocket Control Units, do you just train in Speed Module 1’s and Blue Circuits, or do you train in the Iron Ore, Copper Ore, Coal, and Crude Oil, and make every intermediary product to output the Rocket Control Units?

IMO it is a different challenge to make everything needed by the desired output in an outpost, compared to making everything needed elsewhere and only bringing in the intermediaries needed to directly produce the desired output.

I’ve done both in the same game, actually. Brought in Plate (had a separate smelter elsewhere that trains brought ore to) to make the GC at one outpost. Brought in Plate, Coal, and Crude Oil to make the RC at another outpost. Brought in the same (in greater quantities) in a third outpost to make the BC. On the other hand, I’d ship in the Iron Plate and Green Circuits to make the Green Science, making the Gears, Belts, and Inserters locally.

[Bauer]

Both ways are fine. As long as the build allows to scale up easily.

However, there is one argument for a modular approach when it comes to construction material (iron-plates, steel, green and red chips, maybe also bricks): While your science set-up will pull stuff at a fixed rate, your demand for building materials is quite un-even. E-furnaces, modules, bots, nuclear power buildings, etc. consume lots of resources. If you want your base to handle such demand peaks without compromising the science production, a modular approach for these receipts is much easier because the buffer at the interface (e.g. train station) will smooth out your demand surge. Also: If you decide you need a 20 tier 3 modules per minute procution for a while, you can much easier "just" scale up the chip production facilities.

These considerations brought me to a mixed set-up. I build separate modules for Fe-plates, Cu-plates, steel, green, red and blue chips, bricks and petro-chem (pet, lube, rocket fuel). The science "block" receives these as input but is otherwise a fully integrated monster that turns a lot of that stuff into 2.688k SPM.

[Nosferatu]

An integrated factory, then, would be one that produces several outputs from a shared pool of intermediates: in my case, heaps of green circuits going towards red and blue circuits as well as modules (and thus ultimately to rocket control units). To repeat, I guess the key point is not the shared pool but that there are several outputs which need to be balanced against each other.

I tend to avoid multiple outputs.
Your factory will always be busier on the input side then on the output side (production usually compresses things)
So producing multiple things in the same factory module will bottleneck faster when you go bigger.

Of course if you like the design challenge go for it

[Laie]

It seems to me the question you were really asking is ... when making (for example) Rocket Control Units, do you just train in Speed Module 1’s and Blue Circuits, or do you train in the Iron Ore, Copper Ore, Coal, and Crude Oil, and make every intermediary product to output the Rocket Control Units?

Mostly the latter, I'm bringing in ores and plastic. And that's really not the question: consuming the green circuits at their place of origin seems to be the right thing to do.

These considerations brought me to a mixed set-up. I build separate modules for Fe-plates, Cu-plates, steel, green, red and blue chips, bricks and petro-chem (pet, lube, rocket fuel). The science "block" receives these as input but is otherwise a fully integrated monster that turns a lot of that stuff into 2.688k SPM.

I don't see the wisdom in ferrying around metal plates. I know how much material my green chip assembly is going to need, and build the required amount of smelters on site. I mean, I have to build them somewhere, so why not where the stuff is needed? One might also turn that argument around: the green chip works are rather small compared to the smelters needed to feed it, so why not tack it on right there. That's a lot of copper plates you never need to load on a train ever again.

...said the layman to the farmer Seeing as I'm struggling to produce 1/10th of your output, I probably shouldn't tell you how to run your shop.

However, there is one argument for a modular approach when it comes to construction material (iron-plates, steel, green and red chips, maybe also bricks): While your science set-up will pull stuff at a fixed rate, your demand for building materials is quite un-even. E-furnaces, modules, bots, nuclear power buildings, etc.

Yeah, there's that. I'm trying to disentangle my construction needs from my science production. I have no clear idea how much materials I really need, but suspect that, on average, it isn't all that much. In the end, it all goes through my hands and I certainly do not scale. Still, when I'm on a building spree I'm tossing out tier-3 modules like candy from a carnival float. I hope that large output buffers are going to work for me.

[eradicator]

I don't see the wisdom in ferrying around metal plates.

It all depends on what you're optimizing for. "Integration" is a method, but what goal are you trying to achieve? To me integration is a method to reduce the transfer count of items, and thus reducing the required transport capacity and UPS impact. But if neither is (yet) a concern then having exactly one outpost per recipe should make scaling up much easier because it reduces the need for planning - you don't even have to care about ratios or anything. If you don't have enough ItemA you expand OutpostA and that automatically distributes to anywhere it's needed.

[JimBarracus]

I guess there is no ultimate answer to this.

Sometimes its handier to smelt at the mining outpost and sometimes its easier to smelt at each production site.
There are many things that can affect it like available space in the early stage or things like resource patch layout.

One time I had built a mixed smelter for iron and copper as a central solution.
Other times I had a base resource main bus (just plates) and crafted everything I needed in the block for that science pack.

Not sticking to the same plan over and over gives a lot of variety on every playthrough.

[Bauer]

However, there is one argument for a modular approach when it comes to construction material (iron-plates, steel, green and red chips, maybe also bricks): While your science set-up will pull stuff at a fixed rate, your demand for building materials is quite un-even. E-furnaces, modules, bots, nuclear power buildings, etc.

Yeah, there's that. I'm trying to disentangle my construction needs from my science production. I have no clear idea how much materials I really need, but suspect that, on average, it isn't all that much. In the end, it all goes through my hands and I certainly do not scale. Still, when I'm on a building spree I'm tossing out tier-3 modules like candy from a carnival float. I hope that large output buffers are going to work for me.

Do yourself a favour and go through the math.
You need 205 smelting actions to build an e-furnace (counting steel as 8).
You need 355 smelting actions for a beacon.
You need 3080 smelting actions for a tier 3 module of which you need 4 per e-furnce for a typical 8 beacon setup.
Sum = 12,880 smelting actions
One such furnace will produce 211.5 items/min. Hence, it takes about 60 mins for the smelter to pay for itself.

For comparison:
1 science/min (all 6, with tear 3 modules and beacon) you need 0.838 furnaces. That one furnace produces stuff for 71.6 science in 60 minutes (which isn't soooo much).

If you want 1000 SPM, you need about 840 furnaces. Let's assume that it takes 5 hours to build this setup (from smelter to lab). I leave the exact calculation as an exercise but I guess that the totally required construction material is at least 3x the stuff for the smelter (I didn't take inserters, belts, etc. into account). Hence, you need a whopping

• 3x840x12,880 = about 33 Mio smelting actions.

In order to achieve this in 5 hours, you need more than 500 furnaces. That is almost the size of your science smelter.
If you use your science smelter with the 840 furnaces to build the science setup, it has to run about 3 hours to produce everything. I guess you have to ferry a few metal plates around for this...

PS: Yes, there is a hen-and-egg problem that I blissfully ignore and the ratios (Cu/Fe/..) are also different for the differnet cases. This is more about the general picture.
PPS: The fact that most of the construction material goes into tier 3 modules justifies to separate also the chip production.