Factorio Forums

Since Production modules rank 3 aren't cheap, where should you put them first? Someone surely has already done the math.

skillabstinenz wrote:Since Production modules rank 3 aren't cheap, where should you put them first? Someone surely has already done the math.

The first thing you want is to put there in the machines that use the most raw resources/time (e.g. Electronic Circuits are one of the better choices, as is Sulfuric Acid). Given that Alien artifact collection cannot be automated, imo, the first place to put prod modules is in your Alien science pack assembler as well as the Research labs to maximize research yield (you can get x1.4x1.2=x1.68 yield that way).

This thread has more: viewtopic.php?f=134&t=5705

Edit: Never use prod modules in Pumpjacks, Speed is always better.

I only use them on oil production, iron and copper are abundant enough to not need them (unless you're playing a starvation map but especially then I'd start with oil).

Then it's a toss up... Speaking in terms of levels, the lower level the production (i.e. closest to raw) the better so more products benefit, so putting them in refineries are the obvious choice. However, if you've got 10 refineries but most of your oil is going through a single sulphur chem plant, it's faster and cheaper to make 20% more sulphur than 20% more crude oil products (as you would have to fill ALL your refineries to get 20%). There's probably a way to calculate it but it's complex and involves how many modules you've actually got to spend.

I typically start with places I have fewer chem plants, sulphur, then plastic, multiply up with sulphuric acid and batteries, then fill my refineries and go for cracking last.

Deadly-Bagel wrote:I only use them on oil production, iron and copper are abundant enough to not need them (unless you're playing a starvation map but especially then I'd start with oil).

Then it's a toss up... Speaking in terms of levels, the lower level the production (i.e. closest to raw) the better so more products benefit, so putting them in refineries are the obvious choice. However, if you've got 10 refineries but most of your oil is going through a single sulphur chem plant, it's faster and cheaper to make 20% more sulphur than 20% more crude oil products (as you would have to fill ALL your refineries to get 20%). There's probably a way to calculate it but it's complex and involves how many modules you've actually got to spend.

I typically start with places I have fewer chem plants, sulphur, then plastic, multiply up with sulphuric acid and batteries, then fill my refineries and go for cracking last.

Yes, if you have plenty of individual modules, then working bottom up is a good way to ensure that you're stretching whatever resource you want to stretch the most, since as many processes that use as possible are affected. However, in terms of limited number of modules (this is the more usual scenario unless resources are quite abundant), how to get the best bang for the resource investment has nothing to do w/ where in the chain it is and everything to do with how much raw resource value worth in materials is being used by that process per unit time (the more raw used there, the more impact productivity modules will have there; note that a process which is capable of using lots of raw but isn't running full tilt because you aren't using enough of it or buffering it anymore than you already have isn't going to be as valuable, so that's something to take into account, ex: Sulfuric acid Chem plants are frequently grossly underutilized in smaller bases).

I would have said that it's not a matter of being "closer to raw", but simply a question of how fast the process consumes anything. Despite being as far away from raw as possible, the Rocket Silo is far and away tops, simply because it chugs back expensive components like crazy. But you wouldn't go using prod modules in an engine assembler, simply because the process only uses a few materials, and does so very slowly. Plastic and sulphur drink petrol quickly, also making them good candidates. Whereas smelters are actually kind of medicore, since they normally take nearly 2 seconds to process a single piece of ore.

Frightning wrote:Yes, if you have plenty of individual modules, then working bottom up is a good way to ensure that you're stretching whatever resource you want to stretch the most, since as many processes that use as possible are affected. However, in terms of limited number of modules (this is the more usual scenario unless resources are quite abundant), how to get the best bang for the resource investment has nothing to do w/ where in the chain it is and everything to do with how much raw resource value worth in materials is being used by that process per unit time ...

That's a good point actually. By default you're probably looking at hitting the buildings (specific buildings, not clusters of buildings) with the highest throughput even if you have quite a lot of modules. I mean if half your oil is going through a single sulphur plant then regardless you're better off dropping modules in that first. It will multiply with any modules you then install further back in the chain.

But the moment you start hitting grey areas like half a dozen battery chem plants you're better off going back to do your refineries first and working along the chain of production.

Actually, thinking about it the formula is simple in premise (more difficult to actually calculate though). You need to work out roughly how much petroleum each building is processing, then install modules in the buildings with the highest figures. It becomes a bit wonky when you try to calculate before and during cracking so I'd just assume the petroleum is equal to the product's worth when cracked to petroleum (eg 10 crude oil = 9 petroleum, 3 light oil = 2 petroleum, etc).

The reason I say to work "closer to raw" is because you guarantee you get an extra 20% of everything. When you start installing them further along the chain of production you're no longer getting '20% more oil', sometimes it's pretty clear like with sulphur and sulphuric acid which you only really need one chem plant for each, but if you have as many plastic + battery chem plants as you have refineries then you're better off hitting the refineries first.

Eg if you have 2 plastic plants (processing 40% of all oil), 4 battery plants (processing 50% of all oil) and 6 refineries, you're better off installing productivity modules in the refineries. For six modules, the refineries would multiply 100 oil by 120% for 120 oil, 20 extra, while the chem plants would multiply 90 oil by 120% for 108 oil, 18 extra.

^ Yea, this is why I pointed out that you have to think about actual throughput, not just theoretical maximum throughput. Mathematically, a fully utilized Sulfuric acid plant is an excellent target for Productivity modules, in practice, unless you are producing a lot of Batteries (or goddamn metric fucktons of Processing units), it's not as good as the napkin math would imply.

Frightning wrote:Edit: Never use prod modules in Pumpjacks, Speed is always better.

But 12 beacons?! It‘s 800% vs 670%*1.2=804% yet prodmods cost more to produce, need more power and cause more pollution. So if you really want MAX oil prod is a little bit better.

Frightning wrote:The first thing you want is to put there in the machines that use the most raw resources/time (e.g. Electronic Circuits are one of the better choices, as is Sulfuric Acid). Given that Alien artifact collection cannot be automated, imo, the first place to put prod modules is in your Alien science pack assembler as well as the Research labs to maximize research yield (you can get x1.4x1.2=x1.68 yield that way).

Yeah, it's about most raw resources/time not "working your way up". After all it‘s not the american dream we‘re talking about *g

So I did a little bit of math.
In my world alien science value = 0, got tons of them, already using them in my switches.
Assembly machine 3 only
all the same value 1 iron = 1 copper = 1 coal (= 1 gas /w oil) since oil is „free/endless“ and you can speed mod the pumpjacks
100% production time assumed, multiply by your factor to compensate 50% equals *0,5

Your tips were right, sulfuric acid is by far the best (11,875 unit/s), if you take oil into account.
Then comes green chips(6,25 unit/s), since you always need those, 100% uptime shouldn‘t be an issue.
After that blue chips, then gear wheels and plastic bars if you consider oil (6-5 unit/s). Then blue science.

Worst are batteries, ores and plates (~0,5 unit/s) … then red science and cracking (0,75-1 unit/s). Oil processing isn‘t that bad, but sucks tons of power and speedmods in pumpjacks is way better if you want more oil.

Labs can be nearly the best/worst choice: if you have the max speedbonus, are researching something with all 4 science packs at 15s per research it‘s nearly the best choice. But, if you don‘t have speedboni, are researching something which doesn‘t use all the packs and has a 60 research time it‘s the worst.
They need only 60W and don‘t produce any pollution. So all in all they are a good choice.

Here‘s the „full“ list http://picpaste.com/prodmods-wDjz82OP.png

skillabstinenz wrote:

Frightning wrote:Edit: Never use prod modules in Pumpjacks, Speed is always better.

But 12 beacons?! It‘s 800% vs 670%*1.2=804% yet prodmods cost more to produce, need more power and cause more pollution. So if you really want MAX oil prod is a little bit better.

Good luck managing to fit 12 Beacons around each of your Pumpjacks, given that you have no control over their layout, not to mention that Beacons are far more cost effective the more things you can affect with each Beacon (which further constrains layout significantly).

An upcoming change (maybe it's been done? idk) is that the minimum output of an oil well, ie when it's run "dry", will scale with the original richness of the well.

Skilla, don't forget to divide the production numbers by the number of assemblers you have. Sure, electronic circuits might be your highest produced item but I usually have something like 50-60 assemblers making them. In this case I would be better off targeting the machines that use them first, namely module assemblers.

Assuming everything is t3 we can calculate in recipe speeds without bothering with the assembler speed (as I'm not including chem plants here).

* One Green Circuit assembler consumes 2 Iron Plates and 3 Copper Plates per second.
* One assembler producing T3 Productivity Modules consumes 18 Iron Plates, 33.3 Copper Plates, 6.3 Coal, and 9.9 petroleum per second.

If you had a productivity module, which assembler would you put it in? =P and as you're making productivity modules for this task I can guarantee that assembler will be active for the foreseeable future.

Don't forget that prod modules can't be used on final stage products, including other modules

Hmmm good point, forgot about that. I thought there would be other components that use more than Electronic Circuits but seems only Processing Units come even close, purely because of the insignificant craft time of Electronic Circuits.

Still, with current map gen options oil is much harder to keep a constant flow of than ores so I still rate oil products as far more valuable.

Frightning wrote:Good luck managing to fit 12 Beacons around each of your Pumpjacks, given that you have no control over their layout, not to mention that Beacons are far more cost effective the more things you can affect with each Beacon (which further constrains layout significantly).

Yeah, speedmod is the way to go, there‘s just this one case, the next to nothing oil world

Deadly-Bagel wrote:An upcoming change (maybe it's been done? idk) is that the minimum output of an oil well, ie when it's run "dry", will scale with the original richness of the well.

Maybe there you want to focus on specific wells, too.

The general strategy for more oil is more wells, speedbeacons with 2 or more wells and lastly speedmods in the wells themselves.

Deadly-Bagel wrote:Skilla, don't forget to divide the production numbers by the number of assemblers you have. Sure, electronic circuits might be your highest produced item but I usually have something like 50-60 assemblers making them. In this case I would be better off targeting the machines that use them first, namely module assemblers.

60 asselmbers r3 (no mods) produce 9k/min green chips. My biggest vanilla production I can think of was 4704 (prod modded). If you‘re in for the long run and have that much production power, why not mod? I have just two r3 assemblers for my green chip production atm, both have 4 prodmods r3, one has two beacons (1,75 speed) and the other one has one (1,125 speed). Quite a small base, but I‘m already transitioning to modded production.

Build a cube with 9 assemblers each surrounded with 12 beacons, should be 4*9=36 prod mods and (4*12+16)*2= 128speedmods. This will give you 12,1k/min or about a stack each second *g

skillabstinenz wrote:Build a cube with 9 assemblers each surrounded with 12 beacons, should be 4*9=36 prod mods and (4*12+16)*2= 128speedmods. This will give you 12,1k/min or about a stack each second *g

They wouldn't affect the central assembler, a more efficient layout is to build lines of beacons and assemblers, so that each assembler is affected by 4 beacons per side. The only problem with that layout is getting enough materials to the assemblers to keep it fed, so I opted for logibots in my kilobase, but because I need a non-trivial number of Roboports, I instead went with 1 line of beacons flanked by assembly machines so that each machine was affected by 4 beacons and each beacon (except those on the ends) affect 8 assembly machines (works for Electric furnaces too since they are the same 3x3 size). My kilobase processes over 7k iron+copper plates/min and the entire solid manufacturing area fits on one screen fully zoomed out (note: most of that ore ends up in high tier projects, such as modules, so the number of process stages that ore undergoes is quite high too). That should give you an idea of how insanely efficient that layout can be, and the best part is that it's completely scalable, you can continue to build in that manner to a practically unlimited scale.

Frightning wrote:They wouldn't affect the central assembler, a more efficient layout is to build lines of beacons and assemblers, so that each assembler is affected by 4 beacons per side. The only problem with that layout is getting enough materials to the assemblers to keep it fed, so I opted for logibots in my kilobase, but because I need a non-trivial number of Roboports, I instead went with 1 line of beacons flanked by assembly machines so that each machine was affected by 4 beacons and each beacon (except those on the ends) affect 8 assembly machines (works for Electric furnaces too since they are the same 3x3 size). My kilobase processes over 7k iron+copper plates/min and the entire solid manufacturing area fits on one screen fully zoomed out (note: most of that ore ends up in high tier projects, such as modules, so the number of process stages that ore undergoes is quite high too). That should give you an idea of how insanely efficient that layout can be, and the best part is that it's completely scalable, you can continue to build in that manner to a practically unlimited scale.

Why wouldn‘t they affect the central one?! http://picpaste.com/fac-x7BFOGPn.jpg
The ones at the corners cost full 12 beacons (12*4). The ones between them only 4 extra beacons each (4*4) and the one in the middle is practically free.
This is quite efficient if you want max speed at your assemblies. Hard to suppy though, but what isn‘t at this stage.

Yeah, lines are good too. But what‘s more efficient the beacons on the inside, outside or both?

Yes, the supplying is a nightmare.
I‘ve just this small test setup http://picpaste.com/bufferchests-vP22SS8o.jpg
But I already need buffer chests, because the inserters have a huge delay when picking items from the belts. Already thought about trains but they are too big to fit between the lines. So it‘s just blue belts and stack inserter or is there a better way?

skillabstinenz wrote:

Frightning wrote:They wouldn't affect the central assembler, a more efficient layout is to build lines of beacons and assemblers, so that each assembler is affected by 4 beacons per side. The only problem with that layout is getting enough materials to the assemblers to keep it fed, so I opted for logibots in my kilobase, but because I need a non-trivial number of Roboports, I instead went with 1 line of beacons flanked by assembly machines so that each machine was affected by 4 beacons and each beacon (except those on the ends) affect 8 assembly machines (works for Electric furnaces too since they are the same 3x3 size). My kilobase processes over 7k iron+copper plates/min and the entire solid manufacturing area fits on one screen fully zoomed out (note: most of that ore ends up in high tier projects, such as modules, so the number of process stages that ore undergoes is quite high too). That should give you an idea of how insanely efficient that layout can be, and the best part is that it's completely scalable, you can continue to build in that manner to a practically unlimited scale.

Why wouldn‘t they affect the central one?! http://picpaste.com/fac-x7BFOGPn.jpg
The ones at the corners cost full 12 beacons (12*4). The ones between them only 4 extra beacons each (4*4) and the one in the middle is practically free.
This is quite efficient if you want max speed at your assemblies. Hard to suppy though, but what isn‘t at this stage.

Yeah, lines are good too. But what‘s more efficient the beacons on the inside, outside or both?

Yes, the supplying is a nightmare.
I‘ve just this small test setup http://picpaste.com/bufferchests-vP22SS8o.jpg
But I already need buffer chests, because the inserters have a huge delay when picking items from the belts. Already thought about trains but they are too big to fit between the lines. So it‘s just blue belts and stack inserter or is there a better way?

I misinterpreted your description of it, now that I see it, that works too, but 2/3 of your beacons are only affecting 2 assembly machines (which means much higher fixed costs; note that this is ignoring the corner cases of the boundaries of such a setup), where as the beacon lines setup has every beacon affecting 8 assembly machines and every assembly machine effected by 8 beacons (so fixed costs are lower and you still get significant speedup; +400% with speedmod3s).

Frightning wrote:I misinterpreted your description of it, now that I see it, that works too, but 2/3 of your beacons are only affecting 2 assembly machines (which means much higher fixed costs; note that this is ignoring the corner cases of the boundaries of such a setup), where as the beacon lines setup has every beacon affecting 8 assembly machines and every assembly machine effected by 8 beacons (so fixed costs are lower and you still get significant speedup; +400% with speedmod3s).

Yes, Lines surrounded by beacon lines seems the most efficient way, with beacons being the longer lines especially in the middle and with assemblies.

But how do you supply them? http://picpaste.com/furnace-3TRXq0PB.jpg
This is my small setup for the furnaces. ATM it does 1481 ore to 1777 plates. The full BP should be able to process 2221 ore to 2666 plate each minute.
But it‘s getting really tricky with belts, with only one type in and out. And there should be beacons on the outside too, efficiency wise like you said.

So is it bot only? As I have read that bots are also better for your UPS.

skillabstinenz wrote:

Frightning wrote:I misinterpreted your description of it, now that I see it, that works too, but 2/3 of your beacons are only affecting 2 assembly machines (which means much higher fixed costs; note that this is ignoring the corner cases of the boundaries of such a setup), where as the beacon lines setup has every beacon affecting 8 assembly machines and every assembly machine effected by 8 beacons (so fixed costs are lower and you still get significant speedup; +400% with speedmod3s).

Yes, Lines surrounded by beacon lines seems the most efficient way, with beacons being the longer lines especially in the middle and with assemblies.

But how do you supply them? http://picpaste.com/furnace-3TRXq0PB.jpg
This is my small setup for the furnaces. ATM it does 1481 ore to 1777 plates. The full BP should be able to process 2221 ore to 2666 plate each minute.
But it‘s getting really tricky with belts, with only one type in and out. And there should be beacons on the outside too, efficiency wise like you said.

So is it bot only? As I have read that bots are also better for your UPS.

My kilobase uses bots, but only 1 beacon row per two rows of furnaces/assemblers, with the space inbetween being for roboports (and storage and substations for power distribution). I use inserter-logistic chest to supply them, works quite nicely, that base is processing ~7k plates/min (Iron and Copper) and most of that is going into high tier projects (Rocket, modules etc.) and yet my 3k bots usually have like 1k idle or more.