[SPZ2-5460][0.1.1] Pipe gate/pipe launcher/catcher distribution issues.

[still umbra]

This is situational, because I have plenty of blueprints where the pipe gate does indeed throughput 1800L/min. But in this case, where im making a smart delta mixer, it does not. Whether it is a problem with the distribution or with the pipe gate itself i'm not sure. The pipe gate is fully upgraded and shows to have a throughput of 1800L/min.

The paint consumption after the pipe gate is 4 mixers, each with a single input, thus 1800L/min

Doubling the amount of pipe gates does solve the problem but i need it to be more compact for this application

[0.1.1] Pipe gate doesn't throughput full 1800L/min

[jade gust]

Can you share a BP just for test purposes

[still umbra]

It said i wasn't allowed to send it for some reason

"it contains content blocked by this server"

[daring chasm]

have you tried sending it as a file ?

[daring chasm]

also, does it still happen at regular speed ? the speed modifying options are only debug features and don't run the simulation precisely

[still umbra]

yes it happens at normal speed aswell, i sped it up here for the purposes of letting it settle faster

I'll try sending it in file form

[jade gust]

I would like the whole setup till the belts you measuring.. so we can check and test all content on this vid

[candid escarp]

I've seen this before but with just fluid launcher (no pipe gate). I believe that multiple launchers and pipe gates along a pipe either do not get filled at the same rate or some other odd interaction.

Same thing happens when I do this:

But doesn't happen if I remove all the intenal launchers:

Most people avoid internal fluid launchers in their designs for this reason. Not saying it's not a bug. Just saying that most people desing around it.

[still umbra]

thats odd, but i need atleast some launchers so i can stack it and wiring around wouldn't make it compact enough for my application. It is nice that it can work like this but i shouldn't need to refrain from using launchers to fix it

[candid escarp]

There's ways of making things compact while reducing the fluid launcher count. Might just need a little re-arranging. I have one mixer design with lots of lunchers that works perfect. Something else I attempted with less launchers had throughput issues.

[jade gust]

In this setup, Blue and green draw box has the same speed on the launchers, when i place the gate on the spot, the green draw box has a lower speed on filling compared with the blue box

So i think there is something indeed with the pipe gate

[candid escarp]

Same thing happens when you replace the pipe gate with a launcher+catcher.

[still umbra]

I shouldnt have to rearrange anything because of this, although it is possible. I should just be able to use the pipe gate and launchers with the given throughput without problems

but good to know that there is a work-around

[candid escarp]

I agree. It probably has to do with the fact that pipes have unlimited throughput.

[still umbra]

Could be

[jade gust]

just a quest out curiosity : Why the pipe gate in that location ?

[still umbra]

it seemed a convenient location to me

because those two portions have got to be cut off from one another while making white paint and have to be connected when making another color

[jade gust]

yeah well, i use a smart mixer, and my pipe gates are behind the input, an 1 each pipe... i guess they has an other flow number ?

[candid escarp]

Also partially has to do with the source pipe not being "saturated":

[still umbra]

the input should be sufficient

[candid escarp]

Yes, but it's an interaction with how the launchers fill up and the catchers empty.

I tried this:

The 8 red producers and the two pipe gates alone would get rid of the problem 99% of the time. Still see an occasional gap. Both get rid of the gapes pretty much 100%. It's just a weird interaction. Probably why we have unlimited pipes now.

[jade gust]

even with 1 pipe gate it works then

so conclusions ? Pipe it self and not pipe gate...

[candid escarp]

I still see an occasional gap with that.

Although it's rare.

[jade gust]

Yeah but those gapes only exits if you move aorund 😄

[still umbra]

Im not sure if i can even fit 12 extra launchers for my application

[indigo tree]

what exactly is your application? perhaps we could help you work around the issue for the time being

[still umbra]

thats very nice of you but i think i can fix it by using the blue print from the (nearly) 2x1 delta smart mixer

i was trying to make my own like i always do because i need to fit 4 onto a 2x4 but i can also do that by slightly modifying that blueprint

although i may still be able to make it work by using two pipe gates here

[indigo tree]

from my own understanding, the issue occurs due to the pipe distribution - it appears to fill all buildings evenly, which means that on a moment-by-moment basis, the launchers ahead of the pipe gate may have a total requested input greater than what the pipe can draw from, so the gate ends up not receiving full capacity 100% of the time, due to the launchers ahead of it drawing in bursts rather than a continuous and smooth draw - I believe that one other solution is to make sure any launchers are always used at full capacity, so their draw is constant and not in bursts

[still umbra]

making the launchers always draw 1800L/min would be really inconvient, but thanks for the advice as i was actually doing the opposite

[candid escarp]

Nice explanation. I figured it was something like that but couldn't wrap my noodle around the details.

[rugged turtle]

This is correct. The game tries to load balance all fluid sources and sinks on each pipe network. The pipe gate is both source and sink (and contains a small amount of fluid) and will split one pipe network into two. Adding a pipe gate in this example changes a balanced network into two unbalanced networks.

[gloomy ridge]

Thanks for reporting this issue 🙂
Definitely an issue.
I am trying to simplify your design now for an internal bug report.

[rugged turtle]

Here's an example of it working as expected. The painter on the right gets only a fraction of what the left painter gets. It's not intuitive.

[gloomy ridge]

I need it broken. Not working as expected 😉
Once I remove the mixers the problem stops happening.
I want to get a blueprint down to the minimal configuration with the bug occurring.

[rugged turtle]

That's my point, it is working as expected. Not intuitive, but expected.

[indigo tree]

one fairly minimal setup is having 2 catchers feed a pipe into three launchers, with one of the three consuming at full capacity (1800/m at max upgrades), and the other two consuming at half capacity (meaning they cause non-even draw); the launcher that needs to run at full capacity will fail, because when the other two launchers send a packet and subsequently refill, the instantaneous demand is 3x launcher, but only 2x catcher is supplied, and the one launcher that should be running continuously will bottleneck

[gloomy ridge]

Thanks 🙂

Yeah. That is not great 🙁

[indigo tree]

Because pipes have no volume, there is no ability to absorb small fluctuations like the launchers create, and that causes the issue

One possible way to solve/mitigate it could be a simple "pressure" effect, making devices with a lower fill level get a higher weight when fluid is distributed

Another is having a small buffer volume (say, 1-2 fluid blobs of capacity) that is shared across a pipe segment, and can absorb and spread out the fluctuations

[gloomy ridge]

To me it just feels like the maths is wrong.

[indigo tree]

Well, the maths currently is that if the two half-capacity launchers need filling, then each of the 3 launchers receives 1/3 of the input capacity (starving the launcher that is at full capacity), and then once they are full again, the full-capacity launcher is happy, but half of the input cannot be accepted because it has nowhere to go

[gloomy ridge]

[SPZ2-5460][0.1.1] Pipe gate/pipe launcher/catcher distribution issues.

[gloomy ridge]

I have created an internal ticket created for further investigation.
Thanks everyone for the help.

[still umbra]

👍👍👍

Just wondering, would placing a fluid tank right before the pipe gate help? So it can cushion the fluctuations

Before it is fixed

[indigo tree]

No, because the fluid tank has a limit on how fast it can fill, of 1800/m, so it cannot fill faster than the gate empties it

[still umbra]

Yeah i know but it doesn’t have to fill faster, it just needs to stay the same which it will because the consumption after it is the same as its input, but it should be balance it out because it consumes a constant 1800L/min

In theory atleast

I’ll test it out when im home

[rugged turtle]

Tanks only input fluid if there is a surplus.

[still umbra]

it does in fact not work :( even when the tank is full it will eventually drain despite the input and output supposedly being equal

[still umbra]

this setup (where the 2 red launchers are replaced by pipes) does solve the problem :D

[indigo tree]

Because the instantaneous input would need to be greater than 1800 in order for the tank to refill whilst the output is at 1800

[still umbra]

i wouldnt expect or want it to fill it up i would've wanted it to stay constant (with some fluctuations)

[indigo tree]

But the input is fluctuating with an average of 1800, so sometimes it is less than 1800, and sometimes it is more, and the tank cannot accept an instantaneous input greater than 1800, so it is impossible for it to take in an average of 1800 if the input fluctuates around an average of 1800

[still umbra]

well it has some stored, so it can actually output 1800 for some time (even when it is receiving less) and then make up for that loss of storage when the output is less than 1800 (and its input is 1800)

Assuming the output is balanced and doesnt require more than 1800L/min ofcourse because that is the max it can output

[indigo tree]

But the output is never less than 1800, so it can never "make up for" any drain, as that would require an input of more than 1800

You have exactly 4 mixer inputs (1800/m) after the gate/tank

[still umbra]

I thought that it could fix the distribution and would remove the uneven amount of paint being consumed (the fluctuations)

[still umbra]

Yeah i get that a max of 1800L/min can flow in and that 1800/min is being consumed after

[indigo tree]

You are correct, but that only works if the input capacity of your buffer is sufficiently above the center point/average of the fluctuations; your fluctuations are centered around 1800, which means you would need a buffer input capacity >1800 to smooth the fluctuations

A pair of tanks would be able to fix it, but a single tank cannot

[still umbra]

I thought that if it were to have about 1800 stored, that that would be suffiecient to buffer the output

[still umbra]

yeah i figured but in that case i was better off using two pipe gates

[indigo tree]

No, because for such a buffer to work, you need to be able to accept the full maximum input rate, rather than the average input rate; the buffer cannot be allowed to restrict the flow on its input

[still umbra]

So you mean that in order for that to work the input of the tanks needs to always be 1800L/min?

[indigo tree]

So if you have an average flow of 100, but the instantaneous flow fluctuates between 50 and 150, you need to be able to accept all 150 on the input of the buffer to smooth the fluctuations

[still umbra]

Or a little more than that which the tank cannot input (but what does become available once the launchers in front consume a little less)

[still umbra]

I know that it doesn't work in this application because of the limited output of a fluid tank but in theory that doesnt have to be the case right? because if it flucates evenly bewteen 50 and 150, and the input of your buffer is 100, it should still be able to output the 150 (assuming the buffer can output that which it cannot in the case of the fluid tank) but that will be draining the tank which it can make up for later when the output is 50.

[indigo tree]

For argument's sake, let us say your setup has a flow that fluctuates between 1200 and 2400 (average 1800); you would need a tank that, at minimum, can input at a rate of 2400 and output at 1800, in order to buffer it

[indigo tree]

It is the limited input of the tank that is the problem here

The fluctuation in your case is on the input side of your tank, not the output

[still umbra]

I was assuming it was the limited output which was the problem, since the output fluctuates and the input does not if adding the fluid tank fixes the uneven distribution

[indigo tree]

Nope, the output fluctuating is because of the fluctuating input, and the fluctuating input is caused by the launches before the gate

[indigo tree]

You are correct here though, just swap output and input for your situation

[still umbra]

its starting to dawn on me now

[indigo tree]

because of the limitation of the input rate on the tanks, the 1200-2400 fluctuation becomes 1200-1800 to the tank, and it no longer is an average input of 1800

[still umbra]

I was under the assumption the part after the fluid tank was the problem (which i thought exceeded the 1800L/min output at one given moment because of launchers synchronizing)

And that the output of the fluid tank was the issue because it could never output more than 1800

[indigo tree]

Nope - technically, it does exceed 1800, but this is not an issue, due to how the launchers are laid out (even if instantaneous demand is not met, flow is divided evenly among all 4 branches in exactly the ratio that each branch consumes at)

(This whole thing is why I despise internal fluid launchers in my builds, lol)