Self-balancing Kovarex setup

Circuit-free solutions of basic factory-design to achieve optimal item-throughput.
Involving: Belts (balancers, crossings), Inserters, Chests, Furnaces, Assembling Devices ...
Optimized production chains. Compact design.
Please provide blueprints!
Forum rules
Circuit-free solutions of basic factory-design to achieve optimal item-throughput
Post Reply
Copy link
TBC_x
Burner Inserter
Burner Inserter
Posts: 18
Joined: Sun Sep 10, 2023 5:20 am
Contact:
Contact TBC_x

Self-balancing Kovarex setup

Post by TBC_x »

So I came up with Kovarex setup that uses belt and inserter mechanics to achieve self-balancing.

Now deadlock-free!
Screenshot
We can use to our advantage the way inserters pick items up from the belt, first we need to set up the inserters to feed back to the same centrifuge. When the centrifuge consumes the items as it starts spinning the inserter tries to top it up and it will pick whatever it needs from the closest lane on the belt. In our case that is U-238 on the inner lane, letting any U-235 from the last cycle through. This can be more fine-tuned by overriding stack size on the inserter (not tested).

The yellow belt slows the entire thing down and ensures that the loop won't get stuck as more U-238 is inserted on the belt.

Advantages:
- No circuits or wires whatsoever
- Very simple, you just create a loop and place centrifuges on the inside
- Easy to scale
- Can be extended with beacons pretty easily, as it is based on very simple principles
- No deadlocks once it's full
Disadvantages:
- Takes a long time before it produces U-235

Old design
Last edited by TBC_x on Sun Feb 04, 2024 6:29 pm, edited 1 time in total.
Top
mmmPI
Smart Inserter
Smart Inserter
Posts: 3922
Joined: Mon Jun 20, 2016 6:10 pm
Contact:
Contact mmmPI

Re: Self-balancing Kovarex setup

Post by mmmPI »

TBC_x wrote: ↑Sat Feb 03, 2024 10:13 pm Disadvantages:
- Takes a long time before it produces U-235
The beacons are somewhat in the way but otherwise this could be modified by making only a small loop for the first 2 centrifuge and then use splitter priority so that only when the first small loop is filled with U-235 it would start overflowing on the rest of the centrifuges or on the next little loop for 2 centrifuge, a bit similar to a chain with rings interlockling. To keep it circuit free, but still force some priority on the first centrifuges so that the first 40 U-235 don't end up split amongst many idle centrifuge but all in the first one or 2 to get it to start producing faster. It wouldn't be perfect priority system but would help in that direction in case that is a concern :)

Also i think maybe later you will be tempted to include an input lane for the recycled U-238 from used fuel with priority over the newly mined U-238 but not over the U-238 coming out of the centrifuge, which should be easy to add to your system using the same tricks you used already but may require switching the side for the input belt coming from north.

I like your blueprint, it also doesn't use filter inserter, which is often a choice to help separate materials on different side of belts for outputs, but you found your way without it, keeping it simple/cheap.
Top
TBC_x
Burner Inserter
Burner Inserter
Posts: 18
Joined: Sun Sep 10, 2023 5:20 am
Contact:
Contact TBC_x

Re: Self-balancing Kovarex setup

Post by TBC_x »

The beautiful thing is that you can modify it whatever way you want. I didn't even attempt to optimize it yet. I just build it in my base, put in some U-235 and forget about it. I think each centrifuge will feed up to 4 centrifuges down the line, then the following ones will top up to 80. Haven't really spent too much time figuring out exactly how it works.

Of course you can make more space between the centrifuges and the beacons for more stuff.
Top
Tertius
Smart Inserter
Smart Inserter
Posts: 1024
Joined: Fri Mar 19, 2021 5:58 pm
Contact:
Contact Tertius

Re: Self-balancing Kovarex setup

Post by Tertius »

I had such setups before (I get bald from pulling my hair off from debugging my deadlocked kovarex setups), and unfortunately they all clog at one splitter later when U-235 or U-238 output overflows.

Your setup as well. Consider the u-235 chests are full and U-235 is accumulating at the outer ring. After some hours, the outer ring fills with u-235 and a few u-238 and the centrifuges are stalling with "output full".

Since the centrifuges are full, the inner ring with u-238 will stall.
The outer ring will continue to run, since there is less u-235 than u-238, but only as long as there is space for u-238 in the belt piece after the right splitter. Both lanes of that belt piece will fill with u-238. The left lane from the occasional u-238 from the centrifuge output. As soon as this belt piece is full with u-238, the splitter stalls.
You try to avoid this situation by adding a slower belt piece (great idea!), however if stuff fills fills up, this doesn't help any more.

This will look like this:
Screenshot 2024-02-04 152853.png
Screenshot 2024-02-04 152853.png (957.29 KiB) Viewed 3773 times
And that's the final state. If U-235 is consumed again, the splitter is still clogged with u-238 and the ring stays stalled:
Screenshot 2024-02-04 153202.png
Screenshot 2024-02-04 153202.png (174.19 KiB) Viewed 3773 times
Top
mmmPI
Smart Inserter
Smart Inserter
Posts: 3922
Joined: Mon Jun 20, 2016 6:10 pm
Contact:
Contact mmmPI

Re: Self-balancing Kovarex setup

Post by mmmPI »

Tertius wrote: ↑Sun Feb 04, 2024 2:33 pm I had such setups before (I get bald from pulling my hair off from debugging my deadlocked kovarex setups), and unfortunately they all clog at one splitter later when U-235 or U-238 output overflows.
I think this one has a slight difference that allow a full circle at red belt speed for the bad 238 uranium despite the presence of the yellow belts in the good 235 uranium loop.
It's a modification from the original with which i couldn't reproduce the problem you mentionned, not sure if it's just a matter of buffering the extra "bad" uranium 238 that could be leftover in the array of centrifuge or if it's something else. I have memories of using design based on similar logic that didn't create problems.
seemok.png
seemok.png (243.72 KiB) Viewed 3743 times


Making an interlocking chain to unlock groups of centrifuges progressively is harder than i thought without using several lane of inputs or different full belts for 235 and 238 or filter inserters which would change too much the original creation.
Top
TBC_x
Burner Inserter
Burner Inserter
Posts: 18
Joined: Sun Sep 10, 2023 5:20 am
Contact:
Contact TBC_x

Re: Self-balancing Kovarex setup

Post by TBC_x »

A full centrifuge might output up to 6 U-238. 2 for the current cycle and 4 from the following two cycles depending on inserter stack size.

I've nailed the deadlocking issue down to insufficient buffering of U-238 on the outer lane. The belt between the splitter and the merge might fill up and block the flow of U-235. You need to be able to buffer at least 6 U-238 per centrifuge. The new design should be fine with something over 800 centrifuges. I've updated the BP with the fixed design.

I haven't properly tested the worst case scenario when all the centrifuges output U-238 so the figures above might be wrong, but the design should absorb all excess U-238 on the outer lane.
Top
Tertius
Smart Inserter
Smart Inserter
Posts: 1024
Joined: Fri Mar 19, 2021 5:58 pm
Contact:
Contact Tertius

Re: Self-balancing Kovarex setup

Post by Tertius »

Oh, you used a chest for buffering. I guess this will last longer until it congests than the lifetime of a map.
You added a quite complex belt+inserter setup. I did that too in my designs, however I was never satisfied with the added spaghetti. Not simple enough any more. I resorted to wiring 2 belt pieces in the end, sacrificing the "no circuit" wish.

The generic issue with deadlocks of this kind is the mixed U-235/U-238 lane. As long there is such a mixed lane and a filtering splitter sorting this lane, and an item needs to change to "the other" output due to filtering, and "the other" output is congested, the splitter will block.

With uranium processing, there are 2 such situations: ore refining with mostly U-238 and occasional U-235, and kovarex with mostly U-235 and occasional U-238.
Either you sort the material with filter inserters and never have mixed lanes, or you keep the belt lane after the splitter free, so items can change to the other output internally.
Top
Post Reply

Return to “Mechanical Throughput Magic (circuit-free)”