Factorio Forums

My design uses some tricks from other designs, clever combinators, heap of wires and my own tricks.

Let me show all the features:

1. It uses 40 pcs U235 as moderator per cycle.
2. It supports any combination of modules, include productivity modules.
3. There is a smart stockpile area. On the start of the process, it stores dark rocks to help quickly find more light rocks in ore processing centriguges. Hoarding satus is ON until all Kovarex machines are up and runing. When achieved, it switches the loading to unloading dark stones back on belt. It is made so, the stockpile has the priority over the fresh resource, so the stockpile is consumed ASAP.
4. It autostarts as soon as first batch of 40 light green rocks are available. First Kovarex machine helps fill second one, and so on. More machines you set, faster they activates.
5. Its compact design with two paralel belts of same orientation. Output belts are one light stones, second dark stones. Easy to enlarge or attach other proceses.
6. Its durable against power outage, overflow/underflow (of Input also output) or the user interference. In case the process is somehow messed up anyway ( the users are cheeky) there is a reset Constant Combinator for each machine (just Off and On again) to fix counting.
7. In the storage area, the Constant Combinator let you set required number of light green stones per machine. (deafult 40) For example, when you have prodced enough but wait time of re-loading lowers your efficiency (or charge your OCD), you can set 80 (instead of default 40) and all the machines auto re-calibrates to your bidding.
8. Power switches can disconnect a section, so power consumption is zero when no production is needed. e.g. no useles beacons power consumption.
9. Use of beacons help to start up the process VERY QUICKLY. Production speed is insane. One Kovarex machine non-beaconed, non-moduled, can support about 65 Nuclear power reactors. With all the beacons and production modules it is about 7 times more. e.g. 455 reactors per machine. So this tiny b-print can suport 1000 reactors continuously runing.

Picture in action:

The added info:
- The stack inserterters need to be set to a stack size which can divide 40 with zero modulo and is higher than 2. e.g. allowed stack sizes are 4,5,8,10. Fastest is obviously 10 (used in blueprint), but you migh use lower number according to your research level. Yellow,Filter inserters are ok, dont change them.

The nerd info:
- In previous version game update 17.x the production module feature produced the bonus item after the standard product, always. So when unloading from machine you always get the bonus piece separate from the rest. This behaviour allowed easyer detection of production bonus. In the version game 18.x the production bonus is sometime released sooner so it mixes with standard product, therfore unloading need different treatment. I had to add one combinator and set as fallows: if box's light rock = 1 then output one light rock. Which fixed the unloading process again. (I dont know exactly when it changed, I know the old system worked last october, but this week it doesnt so I guess its a big change reserved for new game version)

Legend:
Light rock >> U-235
Dark rock >> U-238

Regular cycle for 4 stack size: (credit to :dpacbach)
------------------
1. Cycle finishes with no extra U-235, just 41 output.
2. The stack inserter takes 4 things at a time from the centrifuge, but first does the U-238 (two of them).
3. The two U-238 get taken from the lower box to the upper box (good).
4. Then from that point on, the original stack inserter takes four U-235 from the centrifuge into the lower box, and each time, the second stack inserter takes those into the upper box (good).
5. 40/4 = 8 so after eight cycles. At the very end, there will only be one more U-235 left, and then the filter inserter will detect that and output it onto the middle belt.
6. Third stack inserter inputs the two U-238 into the centrifuge, then that causes the inserter at the very top to turn on and retrieve three U-238 from the top belt, starting the next cycle.
7. Success.

Cycle with extra U-235 that appears due to the productivity bonus:
--------------------------------------------------------------------------------------------------
All steps above are the same, except for step 5:

5. At the very end, there will only be two more U-235 left, and then the filter inserter will detect those and output them on the middle belt.

Copy blueprint
Picture of BP:

Edit: fixed the output inserter Override stack from 1 to 2 which caused an issue on some computers. thx to dpacbach

Edit2: changed layout to allow 12 beacons per Kovarex machine. Beacons also speed up the basic uranium ore proceessing so the initial looking for first 40 pcs of U-235 is faster. Once production started, the beacons could be deconstructed ( perhaps used somewhere else). Keep in mind, one Kovarex machine non-beaconed, non-moduled, can support about 65 reactors. With all the beacons and production modules it is about 7 times more. e.g. 455 reactors per machine. So this tiny b-print can supply 1000 reactors runing. Never heard of such gigabase.

Edit3: thanks to kjagodka for tips (I used the added combinator in bottom as a solution to my existing setup, but you re-thing it from ground up so you find up other logic for detection)
- changed logic to detection of productivity modules
- changed beacons layout so 2 less beacons are used (more compact too) but 12 beacons still affects a machine

Hi gGeorg,

I gave this a try and analyzed the combinator logic and it appears to make sense (see my analysis below and tell me if it is correct). However, I was then surprised to see that when I tested it it doesn't appear to work quite right when the extra U-235 gets emitted due to productivity modules. Each time that happens, an extra U-235 accumulates inside the Centrifuge and does not get emitted. I tried to take a screenshot, but the screenshot does not include the open windows showing contents of centrifuge. I started the setup by placing both U-235 and U-238 on the input belt below, then the filtered splitter separated the two and caused the U-235 to go in the middle belt and the U-238 to go to the top belt, so far so good. Then the process started up efficiently and a few cycles went by and the 41st U-235 was correctly output to the middle belt, so far so good. But then after a couple of cycle, when the 42nd U-235 was emitted due to productivity modules, it was not detected properly, and it ended up going back to the boxes and back into the centrifuge. Each time the 42nd U-235 appears, it gets accumulated in the centrifuge, and that quantity gradually grows. I am using the latest version 18.22.

I am actually confused about this, because my analysis showed that the inserter which outputs the U-235 onto the middle belt should output whether there is either 1 or 2 U-235 present, and so I thought the procedure should go like this with your setup:

Regular cycle:
------------------
1. Cycle finishes with no extra U-235, just 41 output.
2. The stack inserter takes 4 things at a time from the centrifuge, but first does the U-235 (two of them).
3. The two U-235 get taken from the right box to the left box (good).
4. Then from that point on, the original stack inserter takes four U-235 from the centrifuge into the right box, and each time, the second stack inserter takes those into the left box (good).
5. At the very end, there will only be one more U-235 left, and then the yellow inserter will detect that and output it onto the middle belt.
6. Third stack inserter inputs the two U-238 into the centrifuge, then that causes the inserter at the very top to turn on and retrieve three U-238 from the top belt, starting the next cycle.
7. Success.

Cycle with extra U-235 that appears at the same time as cycle completion:
--------------------------------------------------------------------------------------------------
All steps above are the same, except for step 5:

5. At the very end, there will only be two more U-235 left, and then the yellow inserter will detect those and output them onto the middle belt. I believe this should happen because the combinator+yellow-inserter at the bottom says "enable yellow inserter to output U-235 if there is either 1 or 2 U-235 in the network".


So therefore I think both cycles should theoretically work properly, but instead I am seeing an extra U-235 accumulate. Perhaps there is some race condition somewhere?

I was able to fix this by making the yellow inserter (the one that puts the U-235 onto the output/middle belt) to have stack size 2 instead of 1, that way it works for both regular cycles and productivity cycles, and no U-235 accumulates in the centrifuge. That said, I still can't figure out why that is necessary to make it work -- your design seems like it should theoretically work in all situations.

dpacbach wrote: ↑Wed May 06, 2020 8:06 pm So therefore I think both cycles should theoretically work properly, but instead I am seeing an extra U-235 accumulate. Perhaps there is some race condition somewhere?

Hello,
thank for your time. I looked at it once more. How many cores your PC has?
There is a chance that your PC process the middle stack inserter sooner than combinator. (box to box)
Here is a fix. : Set Override stack size for yellow output inserter for U-235 from 1 to 2. That remove the option to skip the procees. Limit the final inserter for 1 was the historical reason anyway.

EDIT: Same solution. LOL. Check your cores. I am using an old intel notebook with 4 threads.

Yes as you mentioned, setting the yellow stack inserter to stack size 2 solves the problem. I don't know how to confirm if the number of CPU cores affects this (it might, or perhaps the order in which the inserters are built affects it), but, either way, making the stack size 2 seems to solves the problem, so perhaps just make that change to the blueprint then it will be perfect. I also tested some of the other properties that you claimed and they seem to work properly.

So conclusion thus far is that your blueprint (with stack size fix) is the best

dpacbach wrote: ↑Wed May 06, 2020 8:52 pm if the number of CPU cores affects this (it might, or perhaps the order in which the inserters are built affects it)

You might be right. Some fluid design work diferently when rotate. Game engine process entities in certain order in the chunk. One way or another, its interesting nitpick. Thank you for help. I have updated the main topic b-print.
Here :

dpacbach wrote: ↑Wed May 06, 2020 8:06 pm 2. The stack inserter takes 4 things at a time from the centrifuge, but first does the U-235 (two of them).

i think there should be U-238.
Could I use your description of process in my first post?

Would you mind check/review my tuned 4-leaf nuclear plant?
(or make a contest of better design off it )

gGeorg wrote: ↑Thu May 07, 2020 4:42 am

dpacbach wrote: ↑Wed May 06, 2020 8:06 pm 2. The stack inserter takes 4 things at a time from the centrifuge, but first does the U-235 (two of them).

i think there should be U-238.

Yes you are right.

gGeorg wrote: ↑Thu May 07, 2020 4:42 am
Could I use your description of process in my first post?

Sure. One situation I did not check with your design is the situation when the extra U-235 (from productivity) emerges in the middle of a cycle, but I think based on my understanding of your combinator logic that it should work, because the yellow inserter will only see a single U-235 in its input network and will thus turn on with the help of the bottom combinator.

gGeorg wrote: ↑Thu May 07, 2020 4:42 am Would you mind check/review my tuned 4-leaf nuclear plant?

Sure I can take a look, probably better to put that in a separate post though.

dpacbach wrote: ↑Thu May 07, 2020 1:01 pm Sure. One situation I did not check with your design is the situation when the extra U-235 (from productivity) emerges in the middle of a cycle, but I think based on my understanding of your combinator logic that it should work, because the yellow inserter will only see a single U-235 in its input network and will thus turn on with the help of the bottom combinator.

You mean the result of productivity, the bonus item? If its released, for some reason in the middle of cycle, or just cheeky player want to feed machine to test it, then unloading stack inserter do his job, unload. When the stone comes into the first box, then standard decision logic process. If its 1 or 2pcs of U-235 then yellow inserters activates and belt it. Otherwise, stack inserter take the rule and pass the load to the second box. e.g. if a cheeky player manualy inser 2pcs of U-235 they are belted. If 3 pcs they are re-inserted to the machine. tested.

My auto-start setup requires no combinators and only one circuit wire:

The output from centerfuges comes along the top belt. As each Kovarex assembler wants they will take in U235 and U238 as they can and output all to the top chest. U238 will be put back on the top belt. All U235 is trickled down to the lower chest and fed back into they Kovarex assembler as much as it wants. I set a high threshold for safety for the stack1 limited to allow flow in the long run:



You can disable the second Kovarex assembler until needed by rotating its U235 filter inserter ...

After seeing this post I decided to try to design smart covarex system on my own.

Copy blueprint
I ended up with this design.
Most of the logic happens in green circuit network.
Combinator on the top counts U-235 taken away from centrifuge, once it reaches 40 it resets and allows blue inserter on the bottom to take remaining U-235.
This design isn't perfect tho, if productivity bonus isn't synced with production it will break.

So I decided to learn from author's design and improve on it.

Copy blueprint

Here most of the logic also happen in the green circuit network.
Inserters which take items from the chest, read items in the chest ande inserter which moves items from the centrifuge.
Inserter on the top turns on, when there is at least stack of U-235.
Inserter on the bottom turns on, when there is less than stack of U-235 and it has to be filtered inserter so it won't take U-238 first.
As in author design, it requires stack size to be divisor of 40 greater than 2.

I also shrinked setup a bit and removed 2 beacons, while keeping ease of extending setup sideways and 12 beacons/centrifuge.

Ranger_Aurelien wrote: ↑Sat May 23, 2020 10:54 pm My auto-start setup requires no combinators and only one circuit wire:

As the topic says, 40pcs (U235) per cycle. Your machine need way more to operate, so doesnt fit criteria.

kjagodka wrote: ↑Fri May 29, 2020 12:31 am Inserter on the top turns on, when there is at least stack of U-235.
Inserter on the bottom turns on, when there is less than stack of U-235 and it has to be filtered inserter so it won't take U-238 first.
As in author design, it requires stack size to be divisor of 40 greater than 2.

I also shrinked setup a bit and removed 2 beacons,

Thanks for the tip wit filter inserter.
I used the added combinator in bottom as a solution to my existing setup, but you re-thing it from ground up so you find up other logic for detection. I ve update the top post to use changes.
- changed logic to detection of productivity modules
- changed beacons layout so 2 less beacons are used (more compact too) but 12 beacons still affects a machine.

However, in your design yo have changed input of U238 so they are inserted to the box after that to the machine. I didnt adopt it. This way, you created redundant buffer of material in the box. I prefer direct insert to keep lean production principles.

Edit: Reposted this to a new topic. viewtopic.php?f=208&t=86382

I took a stab at a Kovarex and fuel rod setup... then I found these. I don't have any beacons or mods going, but I'm sure they can be added. It does do auto start at 40 235s, stores the 328 for enrichment and fuel rod creation. I do use circuits to calculate the inserter stack size on the catalyst moves. This allows for using full stack sizes and factors of 40 are not a problem.

I did not design is to be tile-able. Just attach it to the end of you Uranium ore processing and let it go. There is a chest for iron plates that can be manually filled, supplied by belt, or replaced with a requester chest.

Copy blueprint



Edit: Adding a more detailed description of the logic.
There is a constant combinator set to the catalyst size on signal '5'.
All the U-235 comes into the top left chest. Both from the processors, and as the result from the enrichment.
It is all unconditionally move to the center chest.
When the center chest reaches >= '5' pcs of U-235 the middle inserter is signaled to start moving in the catalyst. The number of pieces moved are counted. The stack size of the inserter is set to number of pieces left to insert into the centrifuge. If the number is greater than the inserter is capable of, it will use is max stack size.
Once '5' pcs of U-235 has been moved into the centrifuge, that inserter is disabled and the remaining U-235 is moved into the fuel rod supply chest.
When the center chest reaches zero U-235 the counters are reset and ready to start the next cycle.

The U-238 comes in on the bottom belt. The leftover from the enrichment process is put into this stream as well. There are two chests here as you typically need space for more than 4.8K of U-238 before you reach your catalyst size. The fast inserter on the bottom will shut off when the entry chest has 3K U-238 to make sure there is room for the Korvarex leftover. This same supply is used for the fuel rod creation.

The iron plates needed for the fuel rods can be supplied manaually, belted in, or brought in by robot. Same with the fuel rod output.

I'd be interested to hear anyone's thoughts

kryojenik wrote: ↑Fri Jun 26, 2020 9:32 pm It does do auto start at 40 235s, stores the 328 for enrichment and fuel rod creation. I do use circuits to calculate the inserter stack size on the catalyst moves. This allows for using full stack sizes and factors of 40 are not a problem.
I'd be interested to hear anyone's thoughts

Your design is more of test bed than actual production blueprint. Tileable design and space for beacons are good practice.

Anyway your current design uses:
9 combinators
9 inserters
5 boxes
Compare to the topic design :
2 combinators
6 inserters
2 boxes

Improvement and optimizations would help. Your design counts every batch which could be advantage when changing number of light stones up and down. For this reason you will need more combintors. I am not sure if 9 are realy needed. Might be interesting design when polished to an industrial level.

For full 12 beacon setups I prefer having 80 u235 circulating per machine, so they have no down time.

Optera wrote: ↑Sat Jun 27, 2020 7:54 pm For full 12 beacon setups I prefer having 80 u235 circulating per machine, so they have no down time.

Sure. My design, the first post, support start with 40 then change required number in one constant combinator and here we go system adapt. Altho it has a problem to go back from 80 to 40.
the kryojenik's test bed can be even better at this, becouse he count every batch, so the system could go up or down, as needed.

gGeorg wrote: ↑Sat Jun 27, 2020 2:39 pm

kryojenik wrote: ↑Fri Jun 26, 2020 9:32 pm It does do auto start at 40 235s, stores the 328 for enrichment and fuel rod creation. I do use circuits to calculate the inserter stack size on the catalyst moves. This allows for using full stack sizes and factors of 40 are not a problem.
I'd be interested to hear anyone's thoughts

Your design is more of test bed than actual production blueprint. Tileable design and space for beacons are good practice.

Anyway your current design uses:
9 combinators
9 inserters
5 boxes
Compare to the topic design :
2 combinators
6 inserters
2 boxes

Improvement and optimizations would help. Your design counts every batch which could be advantage when changing number of light stones up and down. For this reason you will need more combintors. I am not sure if 9 are realy needed. Might be interesting design when polished to an industrial level.

Thanks for the feedback. Did some re-engineering and posted in a separate thread. viewtopic.php?f=208&t=86382. Uses less boxes and less combinators (still more than yours though).