Factorio Forums

h.q.droid wrote: Mon Jun 29, 2026 8:25 am Just one small reminder: all the "real" factories you show here will eventually jam when legendary output were constantly consumed. You need circuitry to dispose excessive intermediates from random swing, which I don't find enjoyable to design.

One of the reasons I prefer space casino over any upcycler setup is that it doesn't need clunky intermediate disposal to avoid jamming, its expected jamming time being exponential rather than quadratic.

It's already built in. Clogging by too much intermediates cannot happen, because consumption is always faster than production of these intermediates. There's no intermediate disposal required, it's intermediate production prevention needed. It's a tiny wire from the main belt (read hold all belts) to the supply belt (enable if supply < limit) to keep the main belts from overflowing with too much input. It's one of the situations where a tiny circuit wire and condition is just magic to make a factory run properly without designing a physical layout with a big footprint just to counter a small edge case. Especially with sushi belt loops. Mechanically controlled sushi belt loops are a nightmare. The belts will all unclog on their own. If you stop extracting the legendary products, it will clog everything, and if you resume extracting, everything will unclog on its own.
I do enjoy designing such stuff.

Tertius wrote: Mon Jun 29, 2026 9:21 am It's already built in. Clogging by too much intermediates cannot happen, because consumption is always faster than production of these intermediates. There's no intermediate disposal required, it's intermediate production prevention needed. It's a tiny wire from the main belt (read hold all belts) to the supply belt (enable if supply < limit) to keep the main belts from overflowing with too much input. It's one of the situations where a tiny circuit wire and condition is just magic to make a factory run properly without designing a physical layout with a big footprint just to counter a small edge case. Especially with sushi belt loops. Mechanically controlled sushi belt loops are a nightmare. The belts will all unclog on their own. If you stop extracting the legendary products, it will clog everything, and if you resume extracting, everything will unclog on its own.
I do enjoy designing such stuff.

Take your blue circuit example, say if you stop recycling or producing uncommon blue when there are 50 uncommon red circuits on the belt. What happens the moment you do accumulate 50 uncommon red circuits on the belt while there are only 5 green circuits from the last recycling?

Tertius wrote: Mon Jun 29, 2026 9:21 am

h.q.droid wrote: Mon Jun 29, 2026 8:25 am Just one small reminder: all the "real" factories you show here will eventually jam when legendary output were constantly consumed. You need circuitry to dispose excessive intermediates from random swing, which I don't find enjoyable to design.

One of the reasons I prefer space casino over any upcycler setup is that it doesn't need clunky intermediate disposal to avoid jamming, its expected jamming time being exponential rather than quadratic.

It's already built in. Clogging by too much intermediates cannot happen, because consumption is always faster than production of these intermediates. There's no intermediate disposal required, it's intermediate production prevention needed. It's a tiny wire from the main belt (read hold all belts) to the supply belt (enable if supply < limit) to keep the main belts from overflowing with too much input. It's one of the situations where a tiny circuit wire and condition is just magic to make a factory run properly without designing a physical layout with a big footprint just to counter a small edge case. Especially with sushi belt loops. Mechanically controlled sushi belt loops are a nightmare. The belts will all unclog on their own. If you stop extracting the legendary products, it will clog everything, and if you resume extracting, everything will unclog on its own.
I do enjoy designing such stuff.

He's referring, I think, to a balancing problem, not a throughput problem.

For example, look at that processing unit loop. When you recycle an uncommon processing unit (or a common one and get a single quality bump), you always get 5 uncommon green circuits, but you have a 50-50 chance of getting an uncommon red circuit.

So while, on average, you get 10 times as many green circuits as you do red circuits, as the system runs it's pretty rare for the numbers to line up exactly like that. Instead, after N recycle recipe crafts you get that produce an uncommon output, you should expect the imbalance to be on the order of sqrt(N).

That is, typically, you expect your uncommon chest buffer to contain either 5 * r * sqrt(N) green circuits or 0.5 * r * sqrt(N) red circuits, and on average r is going to be around 0.4. (1 / sqrt(2 * pi)) Maybe I'm off by a factor of 2 or something; I didn't check my work carefully.

(the math: the number of red circuits will follow a binomial distribution; approximate it with a normal distribution and compute the expectation of the absolute value of the difference from the mean)

What this means is that your system is probably going to run fine after thousands of results, but after millions of results you have a decent chance that you have produced 10000 surplus uncommon green circuits without any uncommon red circuits to pair them with, thus filling up your buffers and making the machine halt.

h.q.droid wrote: Mon Jun 29, 2026 9:41 am Take your blue circuit example, say if you stop recycling or producing uncommon blue when there are 50 uncommon red circuits on the belt. What happens the moment you do accumulate 50 uncommon red circuits on the belt while there are only 5 green circuits from the last recycling?

You have a keen eye. That's a crucial part of that production line. There's a chest all uncommon circuits are being put into, and the other qualities as well into their own chests, and the EMPs are being supplied from these chests, so the belt isn't clogging as long as the chest isn't full. The chests are also solving a throughput problem - stack inserters fully work out of the chests with full hands.
And it's actually the only part from all I showed that's not yet circuit-controlled and may actually clog due to the chest getting full, because the normal-quality EMP is the only machine that's producing more than is being consumed (it has 5 recyclers instead of 4) to keep all machines running even if some quality level is being skipped at recycling. There's one wire missing from that chest to the leftmost EMP to stop that EMP if the chest has > x green circuits.

It didn't happen yet the chests got full with green circuits only - until now, recycler result was sufficiently balanced. To make it bullet proof, it's necessary to destroy stuff from these buffers, yes. Since there is already trash removal for many other material, it's just an entry in some constant combinator and changing the steel chest into a buffer chest as far as I see. On the other hand, I guess I just use legendary steel chests here and wait how many years until it clogs.

Tertius wrote: Mon Jun 29, 2026 9:57 am On the other hand, I guess I just use legendary steel chests here and wait how many years until it clogs.

Could even wire up an alarm for manual intervention. "Anything > 9000" for this chest.

Tertius wrote: Mon Jun 29, 2026 9:57 am It didn't happen yet the chests got full with green circuits only - until now, recycler result was sufficiently balanced. To make it bullet proof, it's necessary to destroy stuff from these buffers, yes. Since there is already trash removal for many other material, it's just an entry in some constant combinator and changing the steel chest into a buffer chest as far as I see. On the other hand, I guess I just use legendary steel chests here and wait how many years until it clogs.

See, even you, a person who enjoy this kind of designs, didn't put the intermediate destroying logic there by default. FYI, my legendary tank belt on a similar design clogged in days. Scale legendary production to science level, and the game either becomes overclogged-chest-cleanup, or destroy-everything-everywhere. Keeping space casino behind Asteroid Productivity 26 allowed relatively fun legendary science. The current 2.1.8 release made that... less fun.

BTW, I did like your pipe / LDS designs. With a single input there is no balance issue.

MBas wrote: Mon Jun 29, 2026 8:40 am It was just one example out of many giving you both (copper and iron) base resources and which can be created in cryogenic plant (meaning there is 8 + 4 modules in one loop). Roughly speaking EM plant, biochamber or foundry gives you 2 "free" module slots because its productivity is equivalent of 2 productivity modules.

I never used LDS or Blue Circuits exploit before because it seems almost like a cheating. Now I am unsure how to grasp situation when casino is nerfed but productivity research on recyclable materials stays.

And why I mentioned batteries in the first place? I still think it is one of the best options. These 8 + 4 modules is very powerful and can beat many other factors (sulfur acid consumption, high production time) easily because how the series is exponential against module contribution.

See table in https://wiki.factorio.com/Quality espetially the part "crafts" when it should be obvious how much is cryogenic plant powerful on this.

And since preffered way is to have 8 prod modules in cryogenic plant and leave all legendary modules to recycler, you may speed up cryogenic plant so its just all about speed of recycler.

But again, I am not saying it is total op. But the casino is not either. You can use green circuit then. With 5 prod modules so you have

Green circuits: Loosing 31.25 percent of material with 4 quality modules

My guess noone is using this just becouse there is still more powerful (and broken in my opinion) processing unit production.

Well, I can't believe I overlooked green circuit upcycling for iron plates. That's actually high-throughput.

Now I recall, the main problem with battery is it's very slow to recycle. Being made in batches of 2, the recycling time not-bug made it 2x slower recycling compared to making, and batteries are slow to make in the first place. The higher yield doesn't mean much when throughput is too little compared to available materials.

h.q.droid wrote: Mon Jun 29, 2026 7:57 am
I think they mean battery upcycling for iron plate. I tried that too because the high theoretical yield. It sucked: batteries are impractically slow to make and consume ridiculously large amounts of acid. One of the reasons space casino worked is because they were quite fast: each chunk was worth 400 plates so the average per-plate time cost was quite low.

The same reason for blue chips: it was also too slow for legendary science at +300% productivity with 31% quality. And the byproduct ain't free (in terms of UPS) to dispose either. My base were dumping entire cars of legendary blue chips as holmium byproduct.

For the normal use of base building materials, they work to some extent. They are still a bit slow if you want to build a legendary promethium ship.

You don't put quality on processing units once they hit 300% productivity. So if you did 300% with -20% speed, it's going to look the same as 300% with 125% speed and so on because you just print 4 chips per recipe.

Even then it isn't a free lunch though because the ratio of parts coming back does some interesting things on belts. Technically you can do things to speed up the recyclers with speed modules. Realistically only the chips you successfully upgrade leave that system and the rest stay to jam up belts or short buffers, so if you try and just throw q2L, s2l into a recycler you'll still have to know what you're doing.

It's also a situation where there's practical hidden factors having to do with recycler behavior that you have to account for if you're going to play games. It isn't impossible, and it's the best alternative for doing general supplies. But it definitely is not asteroid rerolling in terms of how you handle the inventory it gives you. I don't mean small random residues I mean at scale.

h.q.droid wrote: Mon Jun 29, 2026 7:57 am I think they mean battery upcycling for iron plate. I tried that too because the high theoretical yield. It sucked: batteries are impractically slow to make and consume ridiculously large amounts of acid. One of the reasons space casino worked is because they were quite fast: each chunk was worth 400 plates so the average per-plate time cost was quite low.

The same reason for blue chips: it was also too slow for legendary science at +300% productivity with 31% quality. And the byproduct ain't free (in terms of UPS) to dispose either. My base were dumping entire cars of legendary blue chips as holmium byproduct.

For the normal use of base building materials, they work to some extent. They are still a bit slow if you want to build a legendary promethium ship.

Eventually it's clarified, my personnal appreciation is that if you want speed you go for upcycling nuclear reactor or EM plant, stuff that represent a lot more material than a module in a space casino would be transforming. For iron plate , i have already discussed with someone trying to convince me to upcycle belt, but it failed , they couldn't convince me away from underground pipes if you want fast "iron plate". Those are different consideration if you want legendary science, or just infrastructure / prom ship , i agree.

I think the point that was made and with which i agree was that there exist different method that in terms of balance complexity/reward were roughly on par with space casino depending on which particular metric you wanted to maximize, like fastest growth, or best UPS , or "simplicity" or "ressource effciency" which appealed to different players making them not necessarily choose space casino all the time. ( henceforth questionning the need to remove them entirely ).

MBas wrote: Mon Jun 29, 2026 8:40 am I never used LDS or Blue Circuits exploit before because it seems almost like a cheating. Now I am unsure how to grasp situation when casino is nerfed but productivity research on recyclable materials stays.

Were you using some form of casino before ? if not was it because you saw it as an exploit similar to LDS/blue circuit, or more something like in your games you didn't see the benefit or using casino ? to me the longer the game the better were space casino because a bit like mining productivity, there was no hard limit on asteroid "chunk" productivity, whereas the other things like LDS and Blue circuit reaches a cap.( and battery or underground pipe are yet another "lower tier"). To me the relation with research is an important factor.This cap was already quite costly to research but also giving more immediate benefit for things outside of quality than asteroid productivity, it made each planets more efficient in various ways. Eventually asteroid productivity was the only infinite science competing with science productivity for me in late/post game. Whereas now i am unsure if other players see reasons to research something else than only science productivity.

Hurkyl wrote: Mon Jun 29, 2026 9:48 am That is, typically, you expect your uncommon chest buffer to contain either 5 * r * sqrt(N) green circuits or 0.5 * r * sqrt(N) red circuits, and on average r is going to be around 0.4. (1 / sqrt(2 * pi)) Maybe I'm off by a factor of 2 or something; I didn't check my work carefully.
(the math: the number of red circuits will follow a binomial distribution; approximate it with a normal distribution and compute the expectation of the absolute value of the difference from the mean)
What this means is that your system is probably going to run fine after thousands of results, but after millions of results you have a decent chance that you have produced 10000 surplus uncommon green circuits without any uncommon red circuits to pair them with, thus filling up your buffers and making the machine halt.

Ah that's some beautiful math in there, all about predicting the necessary buffer for intermediate byproduct of upcycling. Now if you are using chests though , and an alarm to ring when it's full, it's not the full Factorio spirit one could say, and there you would want to automate what happens when the alarm rings. The same challenge design manifest itself when you try to make a space casino, both the throughput and the ratios are to be considered so that the belts are properly dimensionned to carry all the potentially clogging material away from the platform and into space before any jamming could occur. It's quite of a challenge too because in space you can't use robots, which are often use to alleviate the logistic concerns. And chunks only stack by 1 in cargobays making it yet another puzzle to have a buffer of 10000. Space casino are a beauty of math and designs

coffee-factorio wrote: Mon Jun 29, 2026 11:32 am You don't put quality on processing units once they hit 300% productivity. So if you did 300% with -20% speed, it's going to look the same as 300% with 125% speed and so on because you just print 4 chips per recipe.

Even then it isn't a free lunch though because the ratio of parts coming back does some interesting things on belts. Technically you can do things to speed up the recyclers with speed modules. Realistically only the chips you successfully upgrade leave that system and the rest stay to jam up belts or short buffers, so if you try and just throw q2L, s2l into a recycler you'll still have to know what you're doing.

It's also a situation where there's practical hidden factors having to do with recycler behavior that you have to account for if you're going to play games. It isn't impossible, and it's the best alternative for doing general supplies. But it definitely is not asteroid rerolling in terms of how you handle the inventory it gives you. I don't mean small random residues I mean at scale.

Well, I was doing it at processing unit productivity 30 (playing for weeks at Fulgora-only 1M espm did that, even researched 25 -> 30 for my leftover assembly plants), so I got +300% productivity without any module. So I put in quality modules with speed beacons and there was still some positive quality...

Maybe one still should prefer speed modules over quality modules at productivity 30 for throughput. But I stopped optimizing due to the constant clogging and not willing to add de-clogging to every version of the draft.

coffee-factorio wrote: Mon Jun 29, 2026 11:32 am You don't put quality on processing units once they hit 300% productivity. So if you did 300% with -20% speed, it's going to look the same as 300% with 125% speed and so on because you just print 4 chips per recipe.

If you are at 300% productivity, why would you put quality on recyclers and then deal with up to 10 different outputs, instead of mixing speed/quality on assemblers for max uplift flow and dealing with only 4?