Some of those numbers look wrong to me. Here are the uranium-related numbers I have been using.
A uranium miner will produce 15.75 uranium ore/minute. (This is exactly half of what iron and copper miners will produce.)
The uranium processing recipe has a crafting time of 10 seconds, and a centrifuge has a crafting speed of 0.75. A single centrifuge will therefore process 45 uranium ore/minute, and produce (approximately) 0.0315 units of U-235 and 4.4685 units of U-238 per minute. That U-235 number is small enough that it is useful to put it in other terms: A single centrifuge will produce 1.89 units of U-235 per hour, plus quite a lot more U-238.
Thus, you want roughly three miners per centrifuge.
A nuclear reactor has a "cycle" time of 200 seconds. To put that another way, a reactor will consume 18 fuel cells per hour. One unit of U-235 (plus some U-238) makes 10 fuel cells, so we could also say that a reactor effectively consumes 1.8 units of U-235 per hour.
1.8 is a very similar number to 1.89. One centrifuge will support one nuclear reactor, with a little bit left over. It is entirely viable to run many nuclear reactors this way: Three miners and one centrifuge per reactor, with a bunch of boxes to store the excess U-238. You can set this up as soon as you have the blue science needed to research nuclear reactors, and the massive amount of copper needed to build one.
(I recommend doing so! Nuclear power provides a huge amount of power for very little upkeep, and requires way less real estate than solar power.)
Kovarex enrichment changes this situation immensely. Where a single centrifuge processing uranium ore will produce 1.89 U-235 per hour, a single centrifuge running Kovarex enrichment will produce 0.9 U-235 per minute. It does this at a cost of 3 units of U-238, which (if you've been stockpiling it for a long time) you probably have a huge pile of sitting around.
This means that a single centrifuge running Kovarex enrichment, on its own, will produce enough U-235 to power 30 reactors, and it will do this by burning through the huge stock of U-238 that you probably have sitting around anyway.
Closing the loop, there is the fuel reprocessing recipe, which will turn 5 used fuel cells into 3 units of U-238. One centrifuge running this recipe can process the output from 15 reactors.
Eventually, after burning through all that U-238 (which should only take weeks), you will reach a mythical steady-state, in which you produce precisely as much uranium as you need to power your reactor, using the Kovarex process to correct the imbalance in the amount of U-235 and U-238 given by the uranium processing recipe. Calculating these "ideal" ratios is a simple bit of linear algebra: At this point, a single Kovarex centrifuge will support (about) 33 reactors; one uranium processing centrifuge will support just shy of 10 reactors; and one uranium mine will support slightly more than 3 reactors.
EDIT TO ADD: Naturally, these rates and ratios can change with the addition of modules, which I leave as an exercise to the reader. 
I just started a RailWorld setting game in 0.15. Excited to get further in the game than I have before (Blue Science). Going to try to get the achievement for creating a train in less than 90 minutes if possible. We'll see!
Just started my second factory with the new patch. Everything is amazingly smoother now that I have a better idea of how to form a main bus and ratio everything off properly.
Only problem, I'm struggling to progress as the only oil spot nearby is absolutely entrenched with biters. I don't know what I'll do other than slowly creep towers towards each nest?
Ouch!
Thus, you want roughly three miners per centrifuge.
Well, I'm here to tell ya: 8 side-by-side miners don't quite keep 8 centrifuges fully occupied. The first 7 are at full duty cycle, but the 8th only runs at about half speed. I'm not sure where the discrepancy is, but it seems to be very large. (And no, the miners aren't bottlenecked on acid, they've got all they can use. )
What I usually do is run extra-yield modules in the initial centrifuge processing, since uranium is high-value, and power is cheap, and then speed modules in Kovarex centrifuges. That's real important when Kovarex first comes up and you're expanding your stock, since it just takes so damn long otherwise. You can switch back to efficiency modules later, if you want to cut down on power use instead... although, by then, you should have so much you won't really care.
I finally really internalized what the appeal is for rails over belts. The thing with belts is that they take no power to run, and they're a nice steady delivery flow, so I've tended to use them most of the time. What I'm realizing now is that if you're going to get really big, running all those belts by hand gets extremely labor intensive. A rail network lets you use the same rails over and over, for multiple different cargo types. They can also be scaled pretty easily to massive levels of throughput; pulling three cars is only a scoche harder than pulling one. Building an extra six lanes in your stations is one hell of a lot easier than running six parallel belts. (Less total throughput, but a tiny, tiny fraction of the labor and material investment.) And, of course, they match the 'main bus' build concept really well, which is something I'm kinda working on now; I'm using my initial 'science base' as a springboard to build the 'real base', which will be used for rocket production.
Overall, I really like 0.15. They took a long time to get it out the door, but it's better in almost every area. There is a very great deal of thought being invested in the game; the UI is a little weird, but it's got just loads and loads of power. Discoverability is poor, but watching the 'tips and tricks' videos on Youtube can really pay off.
I haven't tried 0.15 yet, because when it dropped I was thoroughly engaged in launching the satellite for the first time. I might be about to try it though, right after I finish getting the "research all techs" achievement.
Of course, that achievement will probably cease to be meaningful since research will now have no end in 0.15?
Only problem, I'm struggling to progress as the only oil spot nearby is absolutely entrenched with biters. I don't know what I'll do other than slowly creep towers towards each nest?
Ouch!
This is one reason I play in peaceful mode. I dont really get how to deal with biter nests until I have a tank and that's pretty late in the game. You can do ok in a jeep but that's a lot of baddies.
The only way I've found to consistently deal with the biters is with laser turrets, and you need petroleum for those, so.....
One real advantage of conveyors over trains: a conveyor belt will not kill you by striking you at 50+ miles per hour.
Not that I have learned this through experience, of course.
Its never not funny though.
I have found a new set of great tutorial/playthrough videos for 0.15. BONUS: the woman's voice sounds similar to Lara Crigger of KaterinLHC/Conference Call fame, and is very nice to listen to.
Playlist (75 videos already!)
The hot early-game shopping mall she uses can be found here: https://factorioprints.com/view/-Kip...
I am going to try it out on my new 0.15 game. I have started a vertical bus like she uses. I also started steel smelting, but noticed that I am having serious power issues due to adding 2 radar stations. So I am learning to use blue prints for the first time to duplicate my steam setup. Very exciting! After I get the power restored, it'll be back to steel smelter, then the mall, then blue science.
I watched some of her tutorials on oil a few months(?) ago and she's fantastic! So happy to see she's doing a series like this. Definitely going to be watching this over dinner for the rest of the week.
Some more thoughts on rails versus conveyors:
1. Conveyors are easy; it's almost the first thing you learn, and you can run them as far as you want to.
2. Conveyors will not kill you by running into you.
3. Rails can scale to arbitrary length; the practical upper limit on conveyor length is probably in the 1000-tile order of magnitude. I've run 1000-length belts, but it's not a lot of fun, and I definitely wouldn't want to go 10K. Rails can probably go for actual miles, as measured on your monitor screen.
4. Conveyors will not flatten you like a pancake for a second time.
5. Rails take a fair bit of overhead; you have to build stations, and you have to think ahead about your station design, at least a little. You'll want to leave expansion room, and you won't even know what that really looks like until you've built a few bases.
6. Conveyors will not murder you like a drunken hobo for a third time.
7. You have to pay attention to your signaling; it's easy to create deadlocks. You end up with something that plays a lot like most train games, in fact, with many of the same puzzles. Building and setting stops, and managing your signals, really has a vibe of Railroad or Transport Tycoon to it.
8. I finally learned to look both ways, and my train death toll remains at three.
Oh, and another mostly unrelated thought: the relatively small change of making underground belts go longer as you tech up? That's actually a big deal. This has really large ripple effects all through the game. As an example, here's a super-efficient design I ran into when looking up how to unload trains faster, in case I hadn't thought of something:
Sure enough, I hadn't thought of something. Not possible prior to 15.7, and suddenly that is a really, really good train station.
edit to add another thing I've noticed: Factorio gets more and more interesting as you scale up. Parallelizing some problems can be quite difficult.
From Kat's videos it looks like they've incorporated several quality of life features that were handled by mods before:
* Ore labels and counts on the map
* Bigger stacks of things like belts
* Upgrades for longer underground tunnels
* Blueprint sharing
* Early blueprinting
* more detailed setup options including more-expensive-everything for a longer ramp and a need to scale everything a bit bigger.
It sounds like early robots are coming too
From Kat's videos it looks like they've incorporated several quality of life features that were handled by mods before:
* Ore labels and counts on the map
* Bigger stacks of things like belts
* Upgrades for longer underground tunnels
* Blueprint sharing
* Early blueprinting
* more detailed setup options including more-expensive-everything for a longer ramp and a need to scale everything a bit bigger.It sounds like early robots are coming too
The early blueprinting and sharing is super nice! Especially with factorioprints.com.
You can get logistic and construction bots and chests with just red & green science, but you need blue science for the personal roboport in your armor.
http://davemcw.com/factorio/tech-tree/
Malor, for train signalling I re-read these tutorials every couple of weeks and grok more every time.
Fell down a Factorio well last night, stayed up past bedtime making green research^^
I can't play Factorio anymore. When Im nearly asleep, my brain dwells on the repetitive placement of belts, altering lines with merging pieces and underground belts. I keep waking up in the middle of the night, I think because I'm annoyed at such inane dreams/near-dreams.
Fell down a Factorio well last night, stayed up past bedtime making green research^^
Yay! I stayed up late working on mine last night, too. I got lube, electric engines, plastic, red circuits and blue science going yesterday. It's nice that blue science doesn't need batteries, so it's a little easier now.
I'm actually cracking oil for the first time in any game, so I'm rather proud of that. About to build my first train outposts because I'm down to about 50k iron ore in my starting deposit. Just need to get all of the gear I will need in my inventory.
Ya, blue was nice to set up...i started on yellow and purple last night and they were a bit more involved. I ended up making a bespoke batch of 150 of each to unlock the full logistics network so hopefully that will make things easier!
Well, there you go. This is definitely a wee-small-hours game. I usually go to bed early these days, but then I started playing Factorio. I guess 2:30 am is "early" in its own way?
Thanks, Mixolyde, for the video link. KatherineOfSky of fantastic! I'm holding off watching too many though, since I don't want to spoil the voyage of discovery as I work my way through my own (terrible, way too small) designs.
Welp, this is the one which has me at the "Stare at the screen and keep forgetting what I was going to do so I'll just randomly run around" level of late night gameplay. After I balked on the game after playing the Demo 6 months ago, or so, I picked it up a few weeks ago and it's sucked me right in.
I have a few false start maps, and am currently working on a map with an ok Bus set up. I think I really screwed up when setting up my oil processing, but I'm working with it for now. Just getting blue up and running.
I just noticed in last friday's blog post the dev reported that Factorio has sold ONE MILLION COPIES. (and they're having a party in Prague to celebrate, inviting players. I'd go if I could!)
This is so good. Very happy for these guys.
I recently started working on my Factorio calculator again, and it's undergone a fairly major overhaul in the process. You can find it here.
The main feature of the calculator is the same as ever: You ask it to calculate the dependencies for one or more items, each at a given rate, and it spits out information about all of the things you need in order to make them. Some of the major new features I've added include:
- Proper handling of "weird" recipes, such as those related to oil and uranium.
- Icons! The calculator is much more graphical now.
- The ability to "ignore" the dependencies of a specific recipe. For example, if you're making red circuits, but you're shipping in green circuits from elsewhere, you can tell it to "ignore" green circuits, and then you'll only see the copper cable requirements for the red circuits themselves.
- Arbitrary numerical precision. Calculations are now performed using bigint rationals, rather than floating-point values, which allows all results to be numerically exact.
- A visualizer, which renders the solution graphically.
- A "nuclear reactor cycle" item, which represents the operation of a nuclear reactor. This allows you to see the exact fuel requirements for a nuclear reactor.
Here are a few examples to get you going:
- How to fuel a nuclear reactor.
- 2000 science per minute, with prod3 modules and beacons across the board.
- Red circuits, with green circuits ignored.
I'm probably forgetting some minor features, and I'm also still cramming new features in there. If anyone has suggestions or comments, I'd be pleased to hear them.
Woah, i loaded up the yellow science. Looking at that chain i can see why it took me so long to get the cycle going! Nice work:)
Nice, MightyMooquack! Love the "DAG".
Nice, MightyMooquack! Love the "DAG".
Technically, it isn't a DAG! Have a look at pretty much any recipe that involves uranium.
Haha. I was just going by the JS library names :D. Lesson learned!
Still grinding away towards my first Satellite launch. Currently playing a peaceful map, so I'll probably turn that off on my next map. Rail signaling really didn't sink in through the in-game tutorials(although I do appreciate them), but I did find one tutorial that dumbed it down enough for me. Chain signal where a train enters a junction, Rail Signal where it exits. If there's another junction directly following, use a Chain Signal instead of a Rail Signal. Currently carting in Iron and Uranium, and supplying the uranium mines with Sulfuric Acid via rail as well.
Now that I'm getting the hang of trains, I'm looking forward to planning out a better system on my next map. I'll stick with my "Just enough to get by for now" design on this map. 
I recently started working on my Factorio calculator again, and it's undergone a fairly major overhaul in the process. You can find it here.
This is insane, I love it!
Now that I'm getting the hang of trains, I'm looking forward to planning out a better system on my next map. I'll stick with my "Just enough to get by for now" design on this map. ;)
There's no particular reason you need to abandon your current map. Almost all long-term players start a map with a "bootstrap" base that gets up to at least blue science. They use that base to build the materials and robots to go build their "real" base nearby, then tear down the old base and re-use the space for something else. With robots, tearing up old stuff is fast and free.
Unless you want to generate a map with different settings like Rail World or something. That's a good reason to start over!
In my game I have two completely separate train systems for iron and stone. I don't need to expand my copper mining for a while. Eventually all the train systems will merge together into a nice system, but for now they are independent and a little messy.
I got construction and logistic bots automatically being added to a roboport, and I'm using a circuit network to control their numbers. I created a small turret blueprint that puts three turrets next to a chest with inserters and a bit of wall around them. Watching the robots repair my turret defenses is pretty cool, but I have to keep an eye on ammo. Next I need to make military science so I can up my weapons tech and take on biter nests. Then I need the next level of modular armor for a personal roboport and other fun modules.
Rail signaling really didn't sink in through the in-game tutorials(although I do appreciate them), but I did find one tutorial that dumbed it down enough for me. Chain signal where a train enters a junction, Rail Signal where it exits.
Basically: use signals to break tracks into chunks; each chunk can have one train on it. Some chunks will have more than one entry path, which means that a train that's in that chunk can block other trains. So, you never want to let a train into an intersection unless it's certain that it can completely exit.
That's what chain signals do: they go green when the next signal is green. In an intersection, that means the exit path has a green signal. So the rule ends up being, chain on entry, regular signals on exit.
This also means that, in NON-intersection chunks with regular signal lights, each chunk should be long enough to contain your longest train.
Unless you want to generate a map with different settings like Rail World or something. That's a good reason to start over!
I just did exactly this, and man, the bootstrap process is so slow. Argh.
I just did exactly this, and man, the bootstrap process is so slow. Argh.
Agreed. Now that I have bots and blueprints, deconstructing and reconstructing the worst parts of my base is SO much easier than starting over.
Pages