Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Power Plants, Energy Storage and Reliable Energy Supply. All about efficient energy production. Turning parts of your factory off. Reliable and self-repairing energy.
Post Reply
Mehve
Filter Inserter
Filter Inserter
Posts: 318
Joined: Sat Aug 06, 2016 9:12 pm
Contact:

Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by Mehve »

Edit: Don't read this first post too hard. A simpler, more compact, and all-around superior iteration was developed further along and can be seen down HERE.

**************************************************

Just like everyone else, been straining my brain, trying to figure out a sleek way of keeping the Kovarex process moving, without tying up too much U235 in the process. My latest attempt seems promising, and has been shooting 100% so far in testing. The only real drawbacks are: 1) requires a minimum of two centrifuges (see bottom), and 2) requires a stack inserter stack size of 7 minimum, and 3) It won't accommodate production modules, if they ever get re-allowed.

The basic premise is this: if you can empty out over half the centrifuges' output in one go, you've got a guaranteed one-time event that you can use for triggering something else. Such as another inserter set to stack size 1. Since each production cycle creates 41 U235, we need to remove at least 21 of them at once. Three stack inserters with a stack of 7+ will do nicely...

Image

And this is basically it. A triple stack inserter unload direct into the next Centrifuge. The lone filter inserter (with stack size forced to one) is set to pull out a single U235 when the three stack inserters are holding their maximum payload (stack size x 3). At the end of the line, simply unload onto a belt that leads back to the start of the line. Also, make sure you're running filtered versions of the inserters and stack inserters, or things foul up right quick.

Image

This pattern can be extended pretty much indefinetely. Current testing is scoring 100% consistency with stack sizes as high as 12, and right down to 7. Since there's only 2 tiles of stuff on each size, feel free to speed beacon the hell out of it.

Image

Blueprints are below. Be aware that, depending on your stack size research level, you'll need to set the constant in EACH filter inserter to be equal to (stack size x 3) - 1.
Minimal Setup Blueprint
Extended Setup Blueprint
Edit: And it just occurred to me that there's no reason you can't run this setup with a single centrifuge. Not sure why I thought otherwise.
Image
Last edited by Mehve on Sun May 28, 2017 9:50 pm, edited 1 time in total.

User avatar
Optera
Smart Inserter
Smart Inserter
Posts: 2915
Joined: Sat Jun 11, 2016 6:41 am
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by Optera »

Great design, simple and efficient.

orzelek
Smart Inserter
Smart Inserter
Posts: 3911
Joined: Fri Apr 03, 2015 10:20 am
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by orzelek »

Very good design - simple and effective :)

quyxkh
Smart Inserter
Smart Inserter
Posts: 1027
Joined: Sun May 08, 2016 9:01 am
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by quyxkh »

Beautiful. I particularly like how clean the U-238 merge is.

Grossen
Burner Inserter
Burner Inserter
Posts: 17
Joined: Tue Oct 11, 2016 5:46 pm
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by Grossen »

Using the design I noticed that when you have stack inserters with a stack size of 8 the blue belt can't handle the flow, Adding a chest buffer before the belt solves the problem trough. Great design :D
Edit: Oh, forgot about the inserter condition xD

Mehve
Filter Inserter
Filter Inserter
Posts: 318
Joined: Sat Aug 06, 2016 9:12 pm
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by Mehve »

Development continues with the "Move over >50% of the outputted U235 at once to establish a one-shot trigger" method... (catchy name still in development)

First up, it occurred to me that you can move over 21 U235 with just two stack filters at 11 or 12 stack size, leaving room for a regular inserter to move the outputted U238 over. This lets you eliminate the U238 return row entirely. For best results, I have the top U238 inserter set to stack size 1, just to make sure it doesn't fill up the target centrifuge first and risk deadlocking the system.
Image
Blueprint
This leads the next logical step - why bother with the main return belt at all? Just have all of them feed directly into the next, in a nice centrifugal circle. This also eliminates the couple seconds of downtime you'd get otherwise, and has the advantage of having the input and output on the same side of the setup, which I tend to prefer wherever possible. Unfortunately, beacon-ability suffers with this pattern, if that's a concern for you.
Image
Blueprint
In an attempt to improve beacon-ability, I reworked the setup into this. With a little belt braiding and a inserter/belt filter to separate the extracted U235, you have a fully packed pattern that allows for maximum beacon efficiency on both sides. Since each centrifuge reuses its own U235, and the input and output are at the same end, you can easily extend the line without interrupting the existing centrifuges.

Image
Blueprint

User avatar
MeduSalem
Smart Inserter
Smart Inserter
Posts: 1475
Joined: Sun Jun 08, 2014 8:13 pm
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by MeduSalem »

I also had an idea on how to improve the circle with direct insertion when I first saw this thread: viewtopic.php?f=202&t=45871
kovarex circle with direct insertion.jpg
kovarex circle with direct insertion.jpg (187.11 KiB) Viewed 17525 times
0.15.6 Blueprint String
My approach detects when the 2 excess U-238 are moved from the previous Centrifuge to the next Centrifuge by one of the Stack inserters. It's a pulse signal on read. This signal triggers the Filter Inserter to output the 1 excess U-235 (with stack size override to 1)... and also pulse-triggers to enter 3 more U-238 into the next centrifuge to make up for the loss.

Beacons are a true problem with those circles because if the speed effect isn't symmetrically distributed over all Centrifuges then the faster ones will eventually stall to wait for the slower ones to catch up.

Also getting the thing started requires manual interference.



But in my Factory I'm currently using following setup:
kovarex 2.jpg
kovarex 2.jpg (147.32 KiB) Viewed 17513 times
0.15.6 Blueprint String
The initial U-235 comes in from the top right (the inserter that is rotated to not do anything)... it will start the process. Then the output U-235 from a Centrifuge is always back-looped to itself until all the Buffer of the Centrifuge is full and once that happens the 1 excess U-235 moves down to the next Centrifuge and rinse and repeat until all Centrifuges are self-sustaining and from then on the excess U-235 moves all the way to the left to the provider chest.

The U-238 part is what I'm proud of... It works by disabling the previous belt tile for as long as the U-238 output filter inserter is holding its 2 U-238. Then when the U-238 input inserter takes 5 U-238 a gap on the belt comes into being and this allows the U-238 output filter inserter to release its 2 U-238 onto the belt, disabling the signal, allowing the belt to fill up again. :D

That setup is beacon-friendly.



I have another one entirely using (Logistic) Chest as feedback loop instead of belts:
kovarex bots+beacon.jpg
kovarex bots+beacon.jpg (281.07 KiB) Viewed 17510 times
0.15.5 Blueprint String
That one works similar to the Ring from above by pulse-triggering the output of the 1 excess U-235 to the provider Chest when the Output Stack Inserter moves the excess U-238 to the Steel Chest.

The Requester Chest is set to request U-238 (to make up for the loss in the process) and also U-235 (to start the process). Once the feedback loop becomes selfsustaining it will not request any further U-235 ever.

The only problem with that one is that one should only use 1 Centrifuge in the beginning and only add one by one after each one is running or otherwise it will take a looong time for all the Centrifuges to start working because of how the Logistic Bots are trying to distribute the initial U-235 equally to the requester chests. So a little bit of manual help might be necessary to get the 40 U-235 together in one centrifuge.

Volley
Manual Inserter
Manual Inserter
Posts: 2
Joined: Fri May 05, 2017 7:54 am
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by Volley »

I have not built it yet, but i have to ask:
Why to use the unloading of U235 as trigger?
The U-238 is a way smaller stack. Wouldn't it be way easier to use the unloading of that as trigger?

milo christiansen
Fast Inserter
Fast Inserter
Posts: 106
Joined: Thu Jul 30, 2015 7:11 pm
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by milo christiansen »

The only reason I can think of would be that U-238 would be more likely to get backed up, failing to trigger properly. Honestly it should work fine as long as you make sure there is always empty belt space to drop the 238.

Mehve
Filter Inserter
Filter Inserter
Posts: 318
Joined: Sat Aug 06, 2016 9:12 pm
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by Mehve »

Volley wrote:I have not built it yet, but i have to ask:
Why to use the unloading of U235 as trigger?
The U-238 is a way smaller stack. Wouldn't it be way easier to use the unloading of that as trigger?
The reason is obviously because... um... I have no idea. Excuse me while I go bang my head on my desk a few times... :oops: Seriously, I have no idea why I decided on that, or why I fixated on it. And in hindsight, the recipe's design makes using the U238 output perfectly viable. And that means that only one inserter is needed to unload all the produced U235, which actually makes a HUGE difference.

Voila. Nothing crazy, but notice the empty spot on top?

Image

Double, extendable voila.
Image

No extra U235 in circulation. It recycles all of its own output, so no loopbacks or interaction/dependencies with any other centrifuges. The only requirements are any inserter capable of moving 2 items in one shot, which even an un-upgraded stack inserter can manage just fine. Only a single circuit wire per centrifuge setup. Yellow belts work just fine if you're on a budget, the only penalty is an extra second or two between cycles.. The twin version can be laid down forever, and the U235 UG sections can be reversed to choose between outputting at the input side or the far side. No need to sort or lane-control the outputted U235. Fully beaconable on both sides and it doesn't matter if the various centrifuges' speeds match or not. The only thing that can jam it is the U235 output belt backing up, which is easy to spot, and can be handled by simply stopping the U238 input.
BP Single Version
BP Extendable Twin Version

User avatar
Optera
Smart Inserter
Smart Inserter
Posts: 2915
Joined: Sat Jun 11, 2016 6:41 am
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by Optera »

Circular builds have 2 major advantages over (most) linear version.
1) they have no downtime cycling material
2) only one centrifuge needs to be filled and the others will eventually also start running

quyxkh
Smart Inserter
Smart Inserter
Posts: 1027
Joined: Sun May 08, 2016 9:01 am
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by quyxkh »

lol if you're going for budget you could even make the front inserters yellow, the payload unloader will never have anything but u235 to choose from. Set it to read/hold, u238=2, extend the wire to the feeback loader set to u235=0.

Mehve
Filter Inserter
Filter Inserter
Posts: 318
Joined: Sat Aug 06, 2016 9:12 pm
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by Mehve »

Optera wrote:Circular builds have 2 major advantages over (most) linear version.
1) they have no downtime cycling material
2) only one centrifuge needs to be filled and the others will eventually also start running
1) True, but even with yellow belts for the recycling part, the downtime between cycling seems pretty minimal. About 10 seconds with yellow belts, only 3-4 seconds with blue belts. Without speed modules, the cycle itself takes over 60 seconds, so I don't worry too much about that. By the time you can afford T3 speed modules, you probably have enough U235 to open up more centrifuges (which cost less than a Speed 3 module), so the individual speed aren't a big concern.
2) Depends on the exact circular build, I suppose. Circular builds with direct insertion avoid the downtime, but I don't think I've seen one that could automatically add the extra U235 into the cycle to get more machines running. They still need each centrifuge to be manually started by adding the first shot of U235, otherwise the centrifuges just going one at a time.

Also, most circular builds don't allow for easy expansion, since you need to reroute belts and/or inserters when you add centrifuges to the mix. Since the centrifuges are completely independent here, you can just keep adding copies to the mix as your U235 stockpiles expand enough to give them the intial U235.
quyxkh wrote:lol if you're going for budget you could even make the front inserters yellow, the payload unloader will never have anything but u235 to choose from. Set it to read/hold, u238=2, extend the wire to the feeback loader set to u235=0.
I hadn't thought about that, but you're right! I could also just use a regular filter inserter instead of the stack version too, since no other unloaded number matters except the U238 unload holding 2 items. Downtime would be high, but I suppose it could be handy if you broke the bank on building the centrifuge and forgot to budget for the inserters :)

Mehve
Filter Inserter
Filter Inserter
Posts: 318
Joined: Sat Aug 06, 2016 9:12 pm
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by Mehve »

@Medusalem

I tend to enjoy running bot-less, (although I've seen you make reference to your potato computer in other threads, so fair) but nice clean designs. Good idea on using the U238 to trigger, as Volley pointed out to me shortly after :) I have no idea how I overlooked that. I've never been entirely clear on how pulse mode works in this game, so I tend to stick to "hold" triggers. And I like that design you currently use. I was mucking around trying to regulate how the new U238 was introduced, in order to guarantee that the recycled U238 would get priority, but I was always trying to work everything at the initial source, rather than use multiple restriction points like you did. Definitely have to keep that in mind for future designs.

Interesting last assembly. Wouldn't that tie up a lot of U235 due to the logistics chest? Not a big deal in the long run, but enough to slow down the initial expansion of U235 stocks?

User avatar
MeduSalem
Smart Inserter
Smart Inserter
Posts: 1475
Joined: Sun Jun 08, 2014 8:13 pm
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by MeduSalem »

Mehve wrote:I tend to enjoy running bot-less, (although I've seen you make reference to your potato computer in other threads, so fair) but nice clean designs.
Yeah, obviously some of the designs (like the circle variant I posted) can be changed to running from belts instead of Chests.

And yes, I'm running on a Core i7 860 that only has 2.8 GHz... With Turbo it climbs to 3.46 GHz... 12 GB RAM DD3 RAM and an embarassing Geforce GTS 250 with 1024MB VRAM. My GTX 570 which I initially used in the computer died off more than a year ago.

With a lot of other games nowadays the CPU doesn't become a bottleneck because they tend to use Dual/Quadcores better, but as in the case of Factorio where the Devs stated multiple times that it doesn't pay off for certain technical reasons (which I'm not quite sure I do understand) it becomes obvious that my CPU seems to be pretty outdated when it comes to Singlecore performance because the game just lags like hell when doing certain stuff in grand scale. 3.46 GHz peak back from 2009 are surely not comparable with what a modern CPU running at the same clock speed is able to do today in 2017.

I know the GTS 250 is also a real bummer, but I only notice that one becoming a bottleneck if I go near to graphically demanding spots... like for example Steam Engines/Turbines with their smoke stuff. There it definitely slows down because of the graphics card... because the slow down goes away when I'm leaving the area.

So eventually I will have to get a new rig anyways, I'm just evaluating when is the best time for it... because if I'm honest... I really miss some innovation from both Intel and AMD (even their Ryzen didn't really get my hopes up).

But that's really off-topic now.
Mehve wrote:Interesting last assembly. Wouldn't that tie up a lot of U235 due to the logistics chest? Not a big deal in the long run, but enough to slow down the initial expansion of U235 stocks?
Yes it does. About 80-90 U-235 in the Centrifuge Buffer, and another 40 + "whatever you set the Requester chest for initial request" in the Requester Chest.... so in total at least 120 U-235 are just there in each Centrifuge to keep it self-sustained.

I know a lot of people would think of that being inefficient, but well... if you have like 10k U-235 already... It doesn't really matter I guess. :D

iceman_1212
Filter Inserter
Filter Inserter
Posts: 256
Joined: Wed Aug 17, 2016 9:49 am
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by iceman_1212 »

i love the elegance of this design. very nice.

and just for fun, here is a botless and almost beltless version that has a centrifuge at crafting speed 6 operating at 100% with 2 cycles worth of input, i.e., system needs to be primed with 80 u235 and 10 u238. when the system gets going, it should show one cycle's worth of inputs in the centrifuge's input buffer (whilst the other set of inputs is being processed.)

there is a similar condition used to input 3 more pieces of u-238 into the system every time the centrifuge outputs 2 pieces of u-238. there are 4 stack inserters at stack size 10 removing u-235, which is ofc massive overkill since all the inserters move exactly once every ~8 seconds or so :lol:

rows of beacons is surely more efficient - this one is just for fun. i also played aroudn with belt setups and noticed that when we have an inserter removing u235 followed immediately by an inserter picking up that u235 off a belt, it helps to have the stack size of the second inserter slightly higher than that of the first.

Image
bp string

iceman_1212
Filter Inserter
Filter Inserter
Posts: 256
Joined: Wed Aug 17, 2016 9:49 am
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by iceman_1212 »

Okay, here is something that is more practical than the abomination in my previous post.

Here is a set-and-forget setup that is intended to be placed in the mid-game right when Kovarex Process research is completed. The beacons are only meant to be added in the late game. Only red/yellow belts are needed as well as the first inserter stack size upgrade.


It has a 99.98% chance of producing the 40 U-235 provided you keep supplying it with Uranium Ore. There is a 0.02% chance of the two buffer chests filling up with U-238 filling up before you get the first 40 U-235.

The setup can operate from a "cold start", i.e., zero U-235. There is one start-up step that's necessary to prime the system: we must insert 5 pieces of U-238 into the centrifuge that will be doing Kovarex Enrichment.

Once the first 40 U-235 is generated, it will spin around in perpetuity at the left-most centrifuge (which does the Kovarex Enrichment Process). The single extra piece of U-235 that results from each cycle of Kovarex is outputted using Mehve's trick that is the core of this thread. In addition, there is one combinator which counts the # of U-235 that is generated from Uranium Processing (i.e., the process which has a 0.7% chance of generating U-235). Once the first 40 are made, it swaps the inserters' settings to send any subsequent U-235 from Uranium Processing to nuclear fuel cell production. (This prevents the centrifuge that's running the Kovarex Enrichment Process from getting backed up due to too much U-235 in it's self-contained loop.)

If you're playing with biters and you want extra U-235 for peacekeeping purposes, then you may need to add centrifuges doing Kovarex Enrichment. Otherwise, once beaconed and moduled, this setup gives ~80 fuel cells per minute which is good for 30+ GW, at which point (if you ever get there) you probably will want to switch to solar for UPS reasons (if you haven't already).

Image

As we can see in the production graph, the system first builds up the 40 necessary U-235. It then initiates Kovarex Enrichment. If/when U-238 gets backed up, it puts backpressure onto Uranium Processing while Kovarex Enrichment continues uninterrupted.
Production Across Stages
Blueprint String

User avatar
brunzenstein
Smart Inserter
Smart Inserter
Posts: 1059
Joined: Tue Mar 01, 2016 2:27 pm
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by brunzenstein »

What is "Mehev's trick"?

iceman_1212
Filter Inserter
Filter Inserter
Posts: 256
Joined: Wed Aug 17, 2016 9:49 am
Contact:

Re: Compact, Extendable, Bufferless, Combi-Free Kovarex Process

Post by iceman_1212 »

brunzenstein wrote:What is "Mehev's trick"?
The idea is to have one filter inserter dedicated to outputting the 1 U-235 of "profit" that is generated each cycle, by limiting its stack size to 1 and by activating it only once at the start of each cycle. Since we know that 2 units of "recycled" U-238 are also generated each cycle, we can use this as our one-time event to activate the first inserter (i.e., the one that's outputting one U-235 which we will extract from the system for use in nuclear fuel cells).

Implementation of this concept:

(1) a filter inserter outputting U-235 set to enabled, stack size 1 and with the condition (U-238 = 2).

...which is connected by a wire to...

(2) a filter inserter outputting U-238 (which will be fed back into the centrifuge), set to None (i.e., it will work unconditionally), stack size 2, Read hand contents and pulse

Optionally, the same method can be used to insert exactly 3 pieces of U-238 at the start of each cycle, which becomes the only input that needs to be supplied to the process.

Among my favorite of the tricks that I've picked up from the forums

Post Reply

Return to “Energy Production”