Need help in math: perfect inserter distribution

[osbornx22]

Now im trying to make satelittes for space science pack. Its the first item, which has

7 inputs:

1 pro uni 100,00
2 low den str 100,00
3 sol pan 100,00
4 acu 100,00
5 roc fue 50,00
6 rad 5,00

and 1 ouput:

7 sat 1

How many input+outputs a assembler can have? max 4*3 sides=12 inserters, right?

from Belt:
Theoretically, we can put just one item of each type on one of the 2 lanes, like:

ore, plate, wheel, ....
engine, x, y, z, ....

So, we can say, just one inserter taking from a belt, can get all kinds of items. Theoretically, we can build everything (not fluid) with just one LANE at the assembler! WTF wow

one belt.jpg (58.67 KiB) Viewed 870 times

I think the same is with boxes, because one box can get input from 3 boxes, which result in endless multiplication.

boxes.jpg (147.15 KiB) Viewed 870 times

I dont know how to make this with circuit network, but i think it should be possible.

But what i really think about is:
Lets say you dont want to do this "everything in one box or one lane " thing.
Lets say you just want to use inserter and belts for input and output.

With 12 beacons and speed modules in the assembler and in the beacons to make satellites is crazy:

speed.jpg (32.21 KiB) Viewed 870 times

This is really impossible or? You can just get 12 inserters to an assembler, but here you need over 40 inserters to feed the assembler fast enough. Crazy.

But it makes sense for satellites, like in real life. Its a big object and you need much stuff. You can put the parts faster together than you can bring in the parts. At some building speed point, the streets around the fabric are too small to feed fast enough. You always have to wait for the next parts in the fabric.

So just forget the beacons and speed modules. Just use efficiency modules (less power need).

Then it looks like this:

effi tool.jpg (76.88 KiB) Viewed 870 times

But i cant place .1 or .6 inserters.
Let say i want to use all 12 normal possible inserters.

How do i have to distribute the input types on each lane, to be proportional to the input needs.

Simple example:
Lets say we need 8 A and 4 B to build 2 C and you can only have 7 inputs.
How do distrubite the 7 possible inputs to the 8+4+2 in/out-comings.
Answer:
A 8/(8+4+2)=8/14=0,57..
B 4/(14)=0,29..
C 2/14=0,14..

A 0,57*7=4 inserter
B 0,29*7=2 inserter
C 0,17*7=1 inserter

Now i know, i cant do more than this. i can try to get more speed, until the inserters dont are fast enough.

The problem is, i tried this for satellites, but i get .x numbers and i dont know how to calculate the perfect distribution:

1 pro uni 100,00
2 low den str 100,00
3 sol pan 100,00
4 acu 100,00
5 roc fue 50,00
6 rad 5,00
7 sat 1

sum = 456
add column with input/sum=ratio

ratio.jpeg (18.54 KiB) Viewed 870 times

now i multiply with max bossible inserter count 12 and round:

runden2.jpeg (31.54 KiB) Viewed 870 times

The problem is, the new sum 13, too much, over 12, thats not possible.

Help

Edit1: The same with 24 lanes:

lanes.jpeg (34.92 KiB) Viewed 867 times

Can use lane number 24 for anything, but then it wont be proportional.
I have an idea: Its not possible to be proportional with this numbers? So there exist no result for the function with "whole numbers" / "integer"?

Edit2: My results makes no sense, because i have 0 inserters for radar or satellite. 1 should be minimum.

Edit3: Sorry, im stupid. Of course there is no result, because if i want to be protortional and the minimum is 1, then i have to set 1 inserter for sattelites and then 100 times more for processing units for example. which direcly result in more then 100 inserters. Im so stupid.

Edit4: Ok forget me, i know nothing. mic drop = give up

[DaveMcW]

You need to build 9 rocket silos to support 9 satellites per minute.

Why not build an extra satellite assembling machine?

[mmmPI]

Now im trying to make satelittes for space science pack. Its the first item, which has

It is not obvious from your calculations that you are aware those quantities required to make 1 satelite could yield a thousand science pack, maybe you did find out with more math .


I think you could consider trains and robots as a way to deliver material too, the first can be used to deliver things in perfect numbers and the second to sidestep the problem ! if you have a train that brings exactly 1 radar which is 20% of the total required, there is enough room in the cargo wagon to place also 20% of all the other required material , or you could have a train that brings exactly enough material to make 1 satellite that's the easiest ! Though you may run into the same problem that if you use the 1-belt-to-deliver-all example in your pictures, that is if you ever have a slight unbalance, like one material running out, then there will be an accumulation on the belt that prevent further required material from arriving and it can get stuck. There will be one train with some radar left or missing and it can get stuck. If you find a way to figure out the logic for the belt problem, there are some other applications for it

Some people feel stupid when they realize something because they think it's so obvious once they do, some others feel very smart, when the exact same thing happens you know? Good luck, have fun !

[Tertius]

I guess from your opening post, this is one of your first satellites you create, may be without having built the rocket silo yet. Or you didn't really look into how long it will take the rocket silo to produce one rocket and how many space science are created by one launch.

A non beaconed and non moduled rocket silo produces one rocket every 100 * 3 seconds = 300 seconds = 5 minutes. This is the time you have for the satellite you want to launch with it. So ignore the actual manufacturing time of 5/1.25 = 4 seconds for one satellite - actually, you have 300 seconds.

If you really produce one satellite every 4 seconds, you need 75 operating rocket silos to consume that. Or 15 fully beaconed rocket silos. 15 fully beaconed rocket silos create about 15 * 1000 = 15000 space science per minute, which is called 15k SPM, which is a really huge megabase one will not be able to build just in a few hours or days.

You probably want 1 rocket silo to do 1 launch per 5 minutes, which is 1000 space science per 5 minutes and requires 1 satellite per 5 minutes. Not 5 seconds.

The challenge with the satellite assembling machine is logistics, but the long distance transport logistics, not the immediate logistics of feeding stuff into the machine. It's more about how and where to produce all those highly refined intermediate products and bring these near the assembling machine.