Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
Forum rules
Circuit-free solutions of basic factory-design to achieve optimal item-throughput
Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
preview
early use case
later use case
Last edited by Theikkru on Fri Oct 14, 2022 4:52 am, edited 10 times in total.
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
What is the circuit programming on the belts for the covarex centrifuges?
Does that prevent the second centrifuge from stealing uranium-235 till the first one is full?
Does that prevent the second centrifuge from stealing uranium-235 till the first one is full?
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
The lower belts enable the upper belt when they have ≥ 10* 235 on them, which means that the first centrifuge is leaving more than a full inserter stack capacity of 235 on the belt. Since the centrifuges only load/unload with 1 inserter, if there's more than the inserter stack capacity on the belt that means there's surplus available for the next centrifuge.
Blueprint and screenshots updated for better recycling setup.
*Edit: after more testing at higher stack capacities I discovered that things can break because the input inserter can be slower than the output inserter due to extending and contracting its arm. I've increased the read belts to 2 and set the threshold to 10, which seems to work: it only starts passing once the first centrifuge is full, and just as importantly, doesn't choke feed to the second centrifuge if there's lots of surplus. It also seems to balance the lanes on the input better, since it forces the inserters to grab more from the far lane than they otherwise would.
Blueprint and screenshots updated for better recycling setup.
*Edit: after more testing at higher stack capacities I discovered that things can break because the input inserter can be slower than the output inserter due to extending and contracting its arm. I've increased the read belts to 2 and set the threshold to 10, which seems to work: it only starts passing once the first centrifuge is full, and just as importantly, doesn't choke feed to the second centrifuge if there's lots of surplus. It also seems to balance the lanes on the input better, since it forces the inserters to grab more from the far lane than they otherwise would.
Last edited by Theikkru on Mon May 27, 2019 7:27 pm, edited 1 time in total.
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
There still seem to be a few odd things in the design.
For example the uranium you get from a left chest is fed to a splitter and then side loaded. But the right chest is not. Remove that and you save 2 underground belts as well.
Next at the bottom the U-235 is side balanced and then side balanced again with a input priority. The covarex process only puts it's output on one side of the belt so the other input doesn't need to be side balanced for the priority splitter.
Actually why use both sides of the belt? How large a setup do you want to build? You could have a single output belt for both types of centrifuges. Processing outputs to the right side of the belt and covarex to the left. At the end a filter splitter sorts U-235 and U-238 and you side feed both onto a single belt. This prioritizes one side of the input belt. Make it so the processing side has priority over the covarex side.
Saves you 4 tiles width, 2 long inserters and a bunch of belts, splitters and underground belts.
For example the uranium you get from a left chest is fed to a splitter and then side loaded. But the right chest is not. Remove that and you save 2 underground belts as well.
Next at the bottom the U-235 is side balanced and then side balanced again with a input priority. The covarex process only puts it's output on one side of the belt so the other input doesn't need to be side balanced for the priority splitter.
Actually why use both sides of the belt? How large a setup do you want to build? You could have a single output belt for both types of centrifuges. Processing outputs to the right side of the belt and covarex to the left. At the end a filter splitter sorts U-235 and U-238 and you side feed both onto a single belt. This prioritizes one side of the input belt. Make it so the processing side has priority over the covarex side.
Saves you 4 tiles width, 2 long inserters and a bunch of belts, splitters and underground belts.
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
In short, the reason for the weirdness you pointed out is that I've designed both Kovarex and processing to produce up to a full belt of product (both lanes). The intent is that once you extend either side (probably Kovarex first) long enough that it's saturating the outer lane, you can add a side-loader so that further extensions feed onto the inner lane. In order for this to work properly, i.e. in order to strictly maintain the priority orders I listed earlier across both lanes, the odd-looking bits have to be included.
For instance, if I didn't side-load the 238 recycling chest, all the recycled 238 would be on the left side of the belt on its way down. As a result, the lane balancer downstream would be receiving recycling only on the top lane, but could take processing product from both lanes. This becomes a problem if the right lane of the bus output is backed up but the left lane is moving, because the processing product would be put onto the left lane before the recycling (violating intended priorities).
The 235 recycling, on the other hand, gets away with only feeding onto one lane, because the feedback loop to Kovarex input will always run faster (by 40× or more) than the output to bus, and the lane that the recycling is feeding onto corresponds to the inner lane on Kovarex input, which input inserters will preferentially grab from.
The 235 loop from processing has a pre-balancer on it for a similar reason. To prevent 235 from backing up and clogging up uranium processing (and possibly starving the system of 238), the 235 from processing must have a strictly higher priority than that from Kovarex so that Kovarex backs up and shuts down first. If the balancer weren't there, it would be possible for 235 from Kovarex to bypass backed up 235 from processing if it were on the opposite lane.
Granted, as you said, all this mess mostly doesn't matter if you only feed products onto the single lanes, but I wanted to future/idiot-proof my design; a few extra yellow belts/undergrounds/splitters are chump change compared to centrifuges and other nuclear apparatus, and a cheap investment to make sure everything just works.
For instance, if I didn't side-load the 238 recycling chest, all the recycled 238 would be on the left side of the belt on its way down. As a result, the lane balancer downstream would be receiving recycling only on the top lane, but could take processing product from both lanes. This becomes a problem if the right lane of the bus output is backed up but the left lane is moving, because the processing product would be put onto the left lane before the recycling (violating intended priorities).
The 235 recycling, on the other hand, gets away with only feeding onto one lane, because the feedback loop to Kovarex input will always run faster (by 40× or more) than the output to bus, and the lane that the recycling is feeding onto corresponds to the inner lane on Kovarex input, which input inserters will preferentially grab from.
The 235 loop from processing has a pre-balancer on it for a similar reason. To prevent 235 from backing up and clogging up uranium processing (and possibly starving the system of 238), the 235 from processing must have a strictly higher priority than that from Kovarex so that Kovarex backs up and shuts down first. If the balancer weren't there, it would be possible for 235 from Kovarex to bypass backed up 235 from processing if it were on the opposite lane.
Granted, as you said, all this mess mostly doesn't matter if you only feed products onto the single lanes, but I wanted to future/idiot-proof my design; a few extra yellow belts/undergrounds/splitters are chump change compared to centrifuges and other nuclear apparatus, and a cheap investment to make sure everything just works.
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
Good plan, too much spaghetti
Unbeaconed:
Beaconed:
Feed ore left: Left bottom chest U238, right bottom chest U235
- Autostart
- Balanced
- Unbeaconed setup produces 5 U235 p/m
- Fully speed-beaconed (centrifuges and beacons) 30 U235 p/m
- Can be starved or backpressure-controlled.
Unbeaconed:
Beaconed:
Feed ore left: Left bottom chest U238, right bottom chest U235
- Autostart
- Balanced
- Unbeaconed setup produces 5 U235 p/m
- Fully speed-beaconed (centrifuges and beacons) 30 U235 p/m
- Can be starved or backpressure-controlled.
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
Spaghetti is cheap (both to build and run). And low tech. Stack/fast inserters, beacons, and level 3 modules are most certainly not. EMBRACE YOUR INNER ITALIAN!
On another note, I'm not seeing how your design gets 238 out without sending it to Kovarex first. What if you need your 238 going to uranium weaponry?
I also see the potential for 235 draining out of Kovarex if a blackout or brownout strikes at an inconvenient time (a side effect of counting on inserters to catch things on a moving belt).
P.S. Statistics! Allow me to join in with some Kovarex costs!
Speed 3 ×2 + Speed 3 beacon ×8: 28+4.8~38.4 MJ/235 (varies depending on beacon sharing), 5⅓ pollution/235
No modules: 21 MJ/235, 4 pollution/235
Efficiency 2 ×2: 4.2 MJ/235, 0.8 pollution/235
(Not including inserter costs)
On another note, I'm not seeing how your design gets 238 out without sending it to Kovarex first. What if you need your 238 going to uranium weaponry?
I also see the potential for 235 draining out of Kovarex if a blackout or brownout strikes at an inconvenient time (a side effect of counting on inserters to catch things on a moving belt).
P.S. Statistics! Allow me to join in with some Kovarex costs!
Speed 3 ×2 + Speed 3 beacon ×8: 28+4.8~38.4 MJ/235 (varies depending on beacon sharing), 5⅓ pollution/235
No modules: 21 MJ/235, 4 pollution/235
Efficiency 2 ×2: 4.2 MJ/235, 0.8 pollution/235
(Not including inserter costs)
Last edited by Theikkru on Tue May 28, 2019 11:17 pm, edited 2 times in total.
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
[EMBRACED]
kovarex consumes almost no 238. the centrifuges produce enough surplus for ammo for the early phase. All 238 is collected in boxes bottom-left. just take out what you need.
You 'll have several thousand 238 before kovarex even starts
https://wiki.factorio.com/Kovarex_enrichment_process
The beaconed version is there as a comparision.
kovarex consumes almost no 238. the centrifuges produce enough surplus for ammo for the early phase. All 238 is collected in boxes bottom-left. just take out what you need.
You 'll have several thousand 238 before kovarex even starts
https://wiki.factorio.com/Kovarex_enrichment_process
The beaconed version is there as a comparision.
Last edited by disentius on Tue May 28, 2019 9:28 pm, edited 1 time in total.
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
I know, but that 238 has a disturbing tendency to evaporate when you realize that shooting and shelling biters is so much more fun with URANIUM!! The case I'm worried about is where your consumption of 238 outstrips production, because with your design Kovarex will be taking its piece of the pie regardless of whether you actually need the 235 or not (unless it's gotten far enough out of hand to back up your 235 line, in which case its consumption was not small by any measure). That's why I put 238 to bus as a higher priority than 238 to Kovarex in my design; if you have a surplus, then fine, Kovarex away, but if you need the 238 for something, Kovarex stops converting it to 235 until you have a surplus again.
Last edited by Theikkru on Tue May 28, 2019 9:32 pm, edited 1 time in total.
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
in that case:
- expand the input ore line and plonk a few more centrifuges down.
- expand the input ore line and plonk a few more centrifuges down.
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
Well that's the true Factorio solution to any problem, obviously, but you have to notice the problem for that to happen, and auto-throttling Kovarex provides a stopgap measure that doesn't drain your dwindling supply in the meantime.
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
You know how much uranium ammo you use? (p/m)
just looked up the numbers, 2 U238/sec uses 24 centrifuges, uses 0.7 red belt ore input, and 0.1 u238 output. works perfect in my setup:)
just looked up the numbers, 2 U238/sec uses 24 centrifuges, uses 0.7 red belt ore input, and 0.1 u238 output. works perfect in my setup:)
Last edited by disentius on Tue May 28, 2019 9:42 pm, edited 1 time in total.
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
And agreed: it is a choice between whichever you need more: fuel for power, or ammo for guns.
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
Ah, yes, I am solidly in the ammo camp, as I run my base defenses on gun turrets and burner inserters, so my effective consumption can be considerably higher than 2×238/s, while conversely, it's only a minor inconvenience if my nuclear plant starves and the lights go out. (Also, there is no "The One Power Pole to Topple it All".)
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
If the U235 is not consumed the kovarex process will fill the belt and eventually halt. And then the output from the normal centrifuges will block at the priority splitter when they produce some U235 and stop working too.disentius wrote: ↑Tue May 28, 2019 7:39 pm Good plan, too much spaghetti
Unbeaconed:
early kovarex1.gif
Feed ore left: Left bottom chest U238, right bottom chest U235
- Autostart
- Balanced
- Unbeaconed setup produces 5 U235 p/m
- Fully speed-beaconed (centrifuges and beacons) 30 U235 p/m
- Can be starved or backpressure-controlled.
You should put the U235 from the normal centrifuges into a chest and use that with priority over kovarex.
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
Good catch mrvn!
I'll see what can be done this weekend. Likely by introducing some spaghetti
I'll see what can be done this weekend. Likely by introducing some spaghetti
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
Next try, already more spaghetti-like
Earlybird version:
Stackable, and upgradeable with beacons, stack inserters, red/blue belts.
Spits out U238 no matter the state of kovarex, and as long as you make sure to have enough storage for U235
Earlybird version:
Stackable, and upgradeable with beacons, stack inserters, red/blue belts.
Spits out U238 no matter the state of kovarex, and as long as you make sure to have enough storage for U235
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
Made one more change to 238 distribution to Kovarex for better efficiency and to handle an obscure case of potential partial backup if the inner lane of Kovarex output gets blocked up by 238 when there is no left lane bus usage of 238 and processing is saturated.
Blueprint and images in the first post have been updated. Also added a second use case screenshot to better showcase how this setup is expandable.
Blueprint and images in the first post have been updated. Also added a second use case screenshot to better showcase how this setup is expandable.
Re: Early-bird's auto-balancing Uranium/Kovarex module for 9-tile bus mount (recycling included)
Redesigned the belt spaghetti to improve prioritization of lanes on the Kovarex side; doing so resulted in an improvement in inserter behavior, and made the 3-belt gating circuit I was using obsolete. I've also switched to feeding 235 output onto a single lane, since the 235 feedback loop will saturate both lanes of a blue belt before output can fill even a single lane of yellow belt.
First post has been updated.
First post has been updated.