3 Belt splitter

[_Ax_]

Hi, I'm setting up a science pack farm and the output it's a 3 lane bus

But I don't know how to split this bus in 3 more bus (to make 3 lanes of laboratories) but maintaining the alignment of the items, like this

Someone know how to do it?

output from the farm

screenshot.PNG (1.45 MiB) Viewed 6243 times

How I want to split the bus

screenshot2.PNG (1.02 MiB) Viewed 6243 times

[Trebor]

1 blue belt = 3 yellow belts.

If you have a fully compressed blue belt you can split it to three fully compressed yellow belts.

This will also let you make a 2 blue to 3 red splitter.

[5thHorseman]

In this situation you can just run each lane through 2 splitters. The first will send half of the packs to one belt and the other half to the second splitter. The other splitter will then split the remaining half into quarters.

This isn't a perfect ratio (of course) but for this case it would likely be good enough. If you want a better ratio, put a third splitter on the lane you never split, and feed the result back into the first splitter's unused input side.

Do this for all 3 belts, and then spaghetti up some undergrounds to get them all in the right places.

[Pandrosos]

You'd need a 1 to 3 balancer on each of the three lanes, then weave all the outputs through to the lab belts. One design is shown here, there are others http://i.imgur.com/l0GhrON.png

Personally though I'd instead make the belts do an S curve to go past all the labs. It'll mean breaking the bottom curve if you want to expand down, but it's so much simpler.

[mrvn]

1) where does space science go? You need 4 belts. Your lines of labs only get 4 of the 6 science packs you have. You need belts on more than one side of the labs.

2) an S or O shaped belt will work nicely. With an O shape you need a ring of labs with and inner and outer ring of 2 belts each so all 7 science packs can be reached by each science lab.

3) I use an + shaped bus. 2 belts going N/S and 2 belts going W/E. Then in each of the 4 quadrants place 3x3 - 6x6 labs with fast and long inserters to feed them from any reachable belts. Between the labs place filter inserters with a stack size override = 1. Set the filters so they only transport science packs perpendicular to the belt that has the science pack. Not along the belt.

4) if you really want a 3-way split then for each belt split the belt, split each output again and loop one of the 4 outputs back as input on the first splitter. Now you get a perfect 1:3 split.

[_Ax_]

1) where does space science go? You need 4 belts. Your lines of labs only get 4 of the 6 science packs you have. You need belts on more than one side of the labs.

2) an S or O shaped belt will work nicely. With an O shape you need a ring of labs with and inner and outer ring of 2 belts each so all 7 science packs can be reached by each science lab.

3) I use an + shaped bus. 2 belts going N/S and 2 belts going W/E. Then in each of the 4 quadrants place 3x3 - 6x6 labs with fast and long inserters to feed them from any reachable belts. Between the labs place filter inserters with a stack size override = 1. Set the filters so they only transport science packs perpendicular to the belt that has the science pack. Not along the belt.

4) if you really want a 3-way split then for each belt split the belt, split each output again and loop one of the 4 outputs back as input on the first splitter. Now you get a perfect 1:3 split.

I have a mod for longer inserters and I've already fixed the problem for the white science pack

[_Ax_]

In this situation you can just run each lane through 2 splitters. The first will send half of the packs to one belt and the other half to the second splitter. The other splitter will then split the remaining half into quarters.

This isn't a perfect ratio (of course) but for this case it would likely be good enough. If you want a better ratio, put a third splitter on the lane you never split, and feed the result back into the first splitter's unused input side.

Do this for all 3 belts, and then spaghetti up some undergrounds to get them all in the right places.

I think I will make that way

[PunPun]

Personally I would not bother with splitting at all and just put inserters between labs so they take from the labs closer to the belt and put into one furter away. You can fit more labs into the space this way and it is also easier.

[MrBadDragon]

Capture.PNG (825.5 KiB) Viewed 6040 times

I like this idea of a Science carousel. It gets round the need for balanced inputs and can feed a large number of research buildings. You can even double side the belts if you need more power!

[slippycheeze]

Personally I would not bother with splitting at all and just put inserters between labs so they take from the labs closer to the belt and put into one furter away. You can fit more labs into the space this way and it is also easier.

FWIW, they will, but a science lab only holds 2 flasks, and wants 2 flasks. So ... once you have a stack of inserter capacity bonus researched (or use a stack inserter) you end up with lab 1 getting science, then lab 2 takes it all out -- and lab 1 stops working -- until lab 1 gets refilled. multiply by the number of labs in sequence, and you can wind up spending more than half your time with labs trying to word and being interrupted by the next one along.

So... this is convenient, but I wouldn't recommend it compared to the belt version, TBH.

Hi, I'm setting up a science pack farm and the output it's a 3 lane bus
But I don't know how to split this bus in 3 more bus (to make 3 lanes of laboratories) but maintaining the alignment of the items, like this

The easy way: you don't have a three lane bus to split, you have three single belt lanes to split. Do each one separately, and use underground links to jump under and over so they get three outputs. Each one is just two splitters plus the original belt for three output lanes. Resynchronize them at the end.

The medium way: don't bother, use circuit connections between the belts to emit a chunk of science from each input to the single output winding past all labs when there are zero bottles of the type that'd be released on the way past. That'll give you a less than fully compressed belts, but it'll work OK for enough labs that you probably don't care.

It'll mess up if you run out of any color. To address that either (a) never let that happen, or (b) have another control that only lets anything trigger if everything has bottles of science waiting.

The hard way: do like the medium way, but use smarter controls to keep a two or three tile long flow of each color pressed against the belt and trying to get in, on both sides of it. That'll ensure as tightly packed a belt as possible. Then have a splitter based filter to extract all the bottles at the end and feed them in on the input line -- preferring them to the new ones from the factory -- so that you automatically fix any balance issue by resorting the far end of the belt to the balanced input system. It'll possibly stall work for a while if you have a missing color, but it will automatically recover as soon as that flows in.

[BlueTemplar]

FWIW, they will, but a science lab only holds 2 flasks, and wants 2 flasks. So ... once you have a stack of inserter capacity bonus researched (or use a stack inserter) you end up with lab 1 getting science, then lab 2 takes it all out -- and lab 1 stops working -- until lab 1 gets refilled. multiply by the number of labs in sequence, and you can wind up spending more than half your time with labs trying to word and being interrupted by the next one along.

So... this is convenient, but I wouldn't recommend it compared to the belt version, TBH.

Between the labs place [...] inserters with a stack size override = 1.

[mrvn]

Two things to add:

1) Note that later techs have mostly 60s per flask. If you don't override the stack size you do get a ripple to refill the science labs but there is still a lot of time spend researching and little refilling if you keep your line of labs reasonably short. The time lost for the next lab grows linear with the number of labs. So it quickly shows diminishing returns.

2) You can put 5-6 labs in the same space as 3-4 with individual belts. So even if you only get 4 out of 6 working on average that's the same research speed.

3) You can place your labs in a square pattern, moving some flasks north/south and others west/east and use up to 3 filter inserter in each direction. That way you can refill 6 science packs in parallel allowing for a larger field of labs without too much downtime. When you reach space science you have to double up something to fit in the 7th pack.

[slippycheeze]

Two things to add:

1) Note that later techs have mostly 60s per flask. If you don't override the stack size you do get a ripple to refill the science labs but there is still a lot of time spend researching and little refilling if you keep your line of labs reasonably short. The time lost for the next lab grows linear with the number of labs. So it quickly shows diminishing returns.

This is absolutely true. It is ironically one place where speeding up inserters or research actually *reduces* per-lab efficiency, but short lines do work.

2) You can put 5-6 labs in the same space as 3-4 with individual belts. So even if you only get 4 out of 6 working on average that's the same research speed.

Yeah. Chasing absolute efficiency is not necessary, and even frequently counterproductive -- efficiency and resilience are on the same line, at opposite ends, so your 100 percent efficient lab setup is also the lab setup that'll grind to a horrible halt because of one science pack out of place. At least, as a rule of thumb.

Personally I see space in Factorio as essentially infinite, so it has zero cost. A slightly wider spaced set of labs, possibly train fed, costs almost nothing more than a super-dense setup. So if you can get high efficiency at the cost of space I'd take that choice.

Ultimately, I build around four lab clusters next to logistic requester chests, and use robots to move the science. Scales better than belts or inserters once you start getting larger fields of the stuff going on. Having a mediocre efficiency layout earlier, or whatever, doesn't really matter because science before logistic chests and robots isn't very expensive, so you don't have too long a wait.

I imagine this differs if you choose, eg, bob, angel, or py mods, which I don't.

[BlueTemplar]

How large is the UPS impact from labs being stopped/restarted by chaining them with stack inserters ? Negligible ?

[mrvn]

How large is the UPS impact from labs being stopped/restarted by chaining them with stack inserters ? Negligible ?

Same cost as every single assembler you have I would guess. So given the ratios of assemblers to science labs that's irrelevant. Also if you set the stack size override to 1 on the inserters the labs don't usually flicker. Only when you take out a flask and the lab uses up a flask at the same time.

Would be nice if faster labs would buffer 3 or 4 recipes worth of flasks.

[Trebor]

Would be nice if faster labs would buffer 3 or 4 recipes worth of flasks.

It would be nice if the labs would not allow inserters to remove the flasks it was using.

[slippycheeze]

Would be nice if faster labs would buffer 3 or 4 recipes worth of flasks.

It would be nice if the labs would not allow inserters to remove the flasks it was using.

It would be nice if the labs would not allow inserters to remove the flasks it was using. ...but at the end of the day, whatever, honestly. I doubt that the UPS cost is significant compared to the inserters, and if you really, really wanted to optimize it, tricking stack inserters into overloading the lab are ====> that way, and don't work with the pass-on model, I think. You would want inserters active as infrequently as possible, because they almost certainly are the highest cost. I'd guess a bot delivery system dedicated to the lab feeding, too, but I'd want to measure before being confident that was cheaper than belt or whatever based input.

[mrvn]

Would be nice if faster labs would buffer 3 or 4 recipes worth of flasks.

It would be nice if the labs would not allow inserters to remove the flasks it was using.

Or at least remove a full flask if more than one is present instead of the partially used flask.