One belt, 2 lane balancer?

[Frost]

Hello, I searched on youtube and cannot find a quick balancer for one belt that remains on one belt, but balances the two lanes on the one belt so that one lane doesn't sit there while the other lane gets used.

If a lane is going left to right, and the bottom lane is not getting used, I can put a splitter so that a new belt shows up above the main belt, and then turn it down to merge with the belt.

That works to a decent degree, but it seems it ends up not being that balanced over time.

So just like they have multiple belt balancers (that I actually use as well and they work fine), can anyone point me to a one belt balancer that keeps both lanes as efficiently balanced as possible? I have yet to figure out one that is as smooth and reliable as any multiple belt balancer.

[Fletch]

Hi @Rio -- perhaps I have terminology backwards (input vs output), I don't want to debate that point.

In fact, I would go so far as to say that the 'simple' 1:1 lane balancer has exactly one purpose: serving as a newb-trap. It doesn't do what the majority of players think it does; and what it actually does do, is 99% useless.

I must disagree. Can you describe the simplest situation that the more complicated lane balancer solves, that the simpler (single splitter) will not?

For OP's use-case, the simpler single-spliter design solves that case perfectly:

balancer for one belt that remains on one belt, but balances the two lanes on the one belt so that one lane doesn't sit there while the other lane gets used.

[RiO]

Originally posted by Fletch:

Hi @Rio -- perhaps I have terminology backwards (input vs output), I don't want to debate that point.

In fact, I would go so far as to say that the 'simple' 1:1 lane balancer has exactly one purpose: serving as a newb-trap. It doesn't do what the majority of players think it does; and what it actually does do, is 99% useless.

I must disagree. Can you describe the simplest situation that the more complicated lane balancer solves, that the simpler (single splitter) will not?

For OP's use-case, the simpler single-spliter design solves that case perfectly:

balancer for one belt that remains on one belt, but balances the two lanes on the one belt so that one lane doesn't sit there while the other lane gets used.

An input balancer balances the ingoing belts or lanes, assuring they are consumed equally.
An output balancer balances the outgoing belts or lanes, assuring they are filled up equally.

OP's target is to "balance the two lanes on the one belt, so that one lane doesn't sit there while the other lane gets used." I.e. their goal is to ensure equal draw, not equal distribution.

They aren't looking to divert one lane to totally different production line than the other lane, and ensuring both lines get 50% of whatever flows in to work with. They want whatever is being consumed to draw 50% evenly on both lanes of the input belt.

What they want is input balancing; not output balancing.


As I said before: the 'simple' lane balancer is a newb-trap. I don't mean that to be offensive to anyone's skill level. I simply mean to say: it is very deceiving in what it actually achieves vs what it doesn't; and unless you're already intimately familiar with how balancers work, it will fool you. At least once or twice.

[Khaylain]

Originally posted by RiO:

Originally posted by Khaylain:

I always prefer the top one in https://i.imgur.com/5rzoobF.jpeg (found by searching google "Factorio lane balancer")

That's a 2:2 lane balancer, but it's not throughput-unlimited. Which means if you use it with only one input belt or only one output belt, it will not properly balance.

If you use only one input belt and only one output belt it's still a 1:1 belt balancer. Not sure how you can say it's not TU, though.

[RiO]

Originally posted by Khaylain:

Originally posted by RiO:

That's a 2:2 lane balancer, but it's not throughput-unlimited. Which means if you use it with only one input belt or only one output belt, it will not properly balance.

If you use only one input belt and only one output belt it's still a 1:1 belt balancer. Not sure how you can say it's not TU, though.

If you use one input belt and one output belt, half the contents won't be lane-balanced and will pass through on the straight bypass before being merged back in. In effect you get one belt with half the contents lane-balanced and the other half as it were, i.e. not balanced.

The single incoming belt will be split 50-50 by the first splitter.
The internal 'simple' 1:1 lane balancer will output balance the contents to evenly distribute over the two lanes 50-50. Then the final splitter will recombine it with the non-balanced other half into one output belt.

Because a 50-50 distributed belt ends up being recombined with a non 50-50 distributed remainder, there will be instances where either lane will temporarily block and in pessimistic cases this can compound to having one of the input lanes temporarily block.

And therefore it is not throughput unlimited.

It's also not universal - because it doesn't actually fully properly balance unless all the inputs and outputs are used and moving. Universality is something practically no balancer has though. Even the basic and well known 4:4 'hourglass' -- which actually is throughput unlimited -- is in fact not universal.

If you would connect up one input belt and three output belts to it, you wouldn't get a 1 -> 0.33 : 0.33 : 0.33 ratio. you'd get a 1 -> 0.25 : 0.25 : 0.5 ratio.

You can make a 4:4 universal balancer as well; but it's at least an approximate 5~6x larger in footprint.

See also: https://alt-f4.blog/ALTF4-27/

[Fletch]

Originally posted by RiO:

....
What they want is input balancing; not output balancing.
....

Sure, let's call it "input" balancing -- I don't want to get hung on up terminology.

As I said before: the 'simple' lane balancer is a newb-trap. I don't mean that to be offensive to anyone's skill level. I simply mean to say: it is very deceiving in what it actually achieves vs what it doesn't; and unless you're already intimately familiar with how balancers work, it will fool you. At least once or twice.

No offense taken. I'm generally interested in learning more. What situation(s) does the more complex lane balancers solve (or solves better) that the simpler one does not handle (or does not handle as well as)? I only use the simpler one and I have quite literally never ran into a case where I think to myself "this isn't working, let me find a more complicated lane balancer design".

All these different lane balancer designs have the same desired effect: filling gaps in an output lane, using excess materials in the other lane.

[Khaylain]

Originally posted by RiO:

Originally posted by Khaylain:

If you use only one input belt and only one output belt it's still a 1:1 belt balancer. Not sure how you can say it's not TU, though.

If you use one input belt and one output belt, half the contents won't be lane-balanced and will pass through on the straight bypass before being merged back in. In effect you get one belt with half the contents lane-balanced and the other half as it were, i.e. not balanced.

The single incoming belt will be split 50-50 by the first splitter.
The internal 'simple' 1:1 lane balancer will output balance the contents to evenly distribute over the two lanes 50-50. Then the final splitter will recombine it with the non-balanced other half into one output belt.

Because a 50-50 distributed belt ends up being recombined with a non 50-50 distributed remainder, there will be instances where either lane will temporarily block and in pessimistic cases this can compound to having one of the input lanes temporarily block.

And therefore it is not throughput unlimited.

-snip for irrelevant-

For the one I said I prefer (for 1:1 belt in:belt out) we know that the first splitter makes it 0.5 each direction. the bottom side shown is essentially switching the side of the lanes (with overflow possibility), then the last splitter merges the direct and the switched lanes belts. This leads to the input being used in 50:50 lane, no matter if the output belt has one lane being unused or both being used in some ratio.

This is also what I've experienced myself, I've experienced no problems with it, and I don't actually see any mathematical reason why it wouldn't work either.

Simply put, I quite believe you misunderstand how it works. Because the point is simply to switch (there's no actual balancing by just switching, you need more than that) on one part and not switch on the other, then merge those.

"The internal 'simple' 1:1 lane balancer will output balance the contents to evenly distribute over the two lanes 50-50." No, it won't. It'll switch the sides, with possible overflow. The "'simple' 1:1 lane balancer" isn't a lane balancer AFAIK, it's a lane switcher with overflow. That's the reason some other designs use undergrounds, to switch sides without any possible overflow. I have found no good reason to actually use undergrounds for this in practice. The only time overflow will matter is if the downstream is managing to pull more than the input is giving, since the other situation where something backs up will lead to one lane per side in the "'simple' 1:1 lane balancer" being backed up always, essentially working like the version with undergrounds that only switch the sides.

I've also used it for 2:2 belt lane balancing too, and the input belts and lanes have also always been pulled from in equal amounts.
It's essentially just a basic splitter (2:2 belt balancer) with some extra within it. And the extra within it does not affect the throughput.

EDIT: Essentially what I'm saying is that if the one I prefer to use isn't TU, then the spliter itself isn't TU either. And I'm pretty sure everyone is going to agree that the splitter itself is TU.

[GAMING_Alligator]

I think all you need for this is a splitter followed by a t-junction.

[Fel]

Originally posted by GAMING_Alligator:

I think all you need for this is a splitter followed by a t-junction.

If you want a "quick and dirty but close enough", then yes, it was even provided with a screenshot in the third answer.
If you want a "proper" balancer you need something a bit more complex but that was also provided in the fourth answer.

[Flash✪Gordon]

Originally posted by Fel:

https://steamcommunity.com/sharedfiles/filedetails/?id=2154178666

EDIT: Maybe words could help, so here is the basic concept.
First you separate the two lanes after using a splitter, usually done through tricks like the sideways underground belts (entrance or exis depending on where you want to send the items).
Then you make sure both belts have the items on the same lane/side and run them through a splitter (the actual "balancing" portion if you will) with 2 inputs and 2 outputs.
Finally you merge them back into 1 belt by forcing each belt to 1 lane of the final belt (since it's already the case for one of them, you usually just turn the other one to go on the other lane).

Why is that better working thatn Fletch's solution?
Without items on the belt the answer is not obvious for me

I followed your explanation till "sideways underground belts" ... there I asked myself why underground is necessary and lost you.

[GAMING_Alligator]

Originally posted by Fel:

Originally posted by GAMING_Alligator:

I think all you need for this is a splitter followed by a t-junction.

If you want a "quick and dirty but close enough", then yes, it was even provided with a screenshot in the third answer.
If you want a "proper" balancer you need something a bit more complex but that was also provided in the fourth answer.

I'm not seeing what's "quick and dirty" about it. It does the same as what you're calling the "proper" solution, but in a much simpler way, without needless additional underground belts and splitters.

If the items are only ever going to be on one side of the input belt, you're going to end up with the same amount of items on each side of the belt once it gets rejoined either way. All that making it complicated does is introduce a bigger possibility of it going wrong later.

[Flash✪Gordon]

Originally posted by GAMING_Alligator:

I'm not seeing what's "quick and dirty" about it. The "proper" solution does the same thing, just in a more complicated way with needless additional underground belts and splitters.

Ha! Thank you for that Mr. Alligator (know you from Satisfactory forum, right?)

I am still beginner but assumed that this underground belts are superfluous.

[GAMING_Alligator]

Originally posted by Gordon✪Gekko:

Originally posted by GAMING_Alligator:

I'm not seeing what's "quick and dirty" about it. The "proper" solution does the same thing, just in a more complicated way with needless additional underground belts and splitters.

Ha! Thank you for that.

I am still beginner but thought that this underground belts are superfluous.

If a problem in Factorio has a simple solution, never underestimate the community's ability to recommend an overly complicated one.

[Khaylain]

Originally posted by Gordon✪Gekko:

Originally posted by Fel:

https://steamcommunity.com/sharedfiles/filedetails/?id=2154178666

EDIT: Maybe words could help, so here is the basic concept.
First you separate the two lanes after using a splitter, usually done through tricks like the sideways underground belts (entrance or exis depending on where you want to send the items).
Then you make sure both belts have the items on the same lane/side and run them through a splitter (the actual "balancing" portion if you will) with 2 inputs and 2 outputs.
Finally you merge them back into 1 belt by forcing each belt to 1 lane of the final belt (since it's already the case for one of them, you usually just turn the other one to go on the other lane).

Why is that better working than Fletch's solution?
Without items on the belt the answer is not obvious for me

I followed your explanation till "sideways underground belts" ... there I asked myself why underground is necessary and lost you.

The one given by Fletch in

Originally posted by Fletch:

-snip
https://steamcommunity.com/sharedfiles/filedetails/?id=3411231230
-snip-

is basically a lane switcher with overflow (unless only having one lane input, since then the overflow on one side becomes an important part to making the output belt have 2 lanes with items).

The one shared by Fel uses the fact that undergrounds block one lane to control which lane goes where, then merge those two back together. The top lane goes to the top belt, but on the bottom lane, the bottom lane (of the input) goes to the bottom belt and also on the bottom lane. Then the splitter puts half of the input (which is 1 full belt worth of items only on the bottom lane) straight to the right on the bottom lane, and half of the input to the top of the splitter output, whereupon it is placed on the top lane of the output.
Thus no matter the input ratio the output of the one Fel showed will give equally distributed output. If the input is less than max and the output has one lane backed up it will also draw equally from both input lanes onto the free output lane.

[Fel]

Originally posted by GAMING_Alligator:

I'm not seeing what's "quick and dirty" about it. It does the same as what you're calling the "proper" solution, but in a much simpler way, without needless additional underground belts and splitters.

If the items are only ever going to be on one side of the input belt, you're going to end up with the same amount of items on each side of the belt once it gets rejoined either way. All that making it complicated does is introduce a bigger possibility of it going wrong later.

It is "quick and dirty" because it doesn't really completely balance things.
It is definitely good enough if all you want is to balance the output, but it isn't balancing the input.
What that means is that if the right lane is the only one really moving after the balancer, it will mean that only the left lane will move before the balancer.
And in many cases, that's perfectly fine, which is why I called it "quick and dirty but close enough", most people including me use this all the time too because it's small in terms of space, very easy to make by hand, and it fits the purpose well enough to not really need something more complec.

But there are cases where you want both lanes before the balancer to be "consumed" at the same speed (assuming both are full or coming at the same rate of course) regardless of what happens after the balancer (even if only one lane ends up moving for example).
In those cases, you will want to use the more complex one.
But since it is a bit more complex, you are probably going to want to use a blueprint (or make it part of a book) for it to make sure you don't make mistakes when building it, which is the same as when making belt balancers really.

[königplatzen]

Originally posted by RiO:

OP's target is to "balance the two lanes on the one belt, so that one lane doesn't sit there while the other lane gets used." I.e. their goal is to ensure equal draw, not equal distribution.

I would come to the same conclusion.

The complex lane balancer pulls equally from both lanes and outputs equally to both lanes.
And, if you consume items on the output, but take from one side here, the input still pulls both sides equally.