Updated Balancers?

[Silvador]

Does anyone know of an up-to-date list of balancer designs? I looked on Google, but everything I found was from five to eight years ago, and I know some older designs have been improved and optimised over the years (and others just plain didn't work properly); and now splitters can prioritise input and output, too? That has to change things. But, I don't have the mental gymnastics to make my own balancers. :c

[Khaylain]

Balancers really don't need the splitters prioritizing input and output, as far as I'm aware. In fact, I'm pretty sure using prioritization on splitters would generally mess up them working as balancers.

https://www.factorio.school/view/-ML5RsMXhj7tnbbzs02H comes up as one of the first ones for me, and that's not 5 to 8 years old.

[Sagittarius_A*]

https://factoriobin.com/post/U5kFRudO

3 sec google search

[Silvador]

Originally posted by Sagittarius_A*:

https://factoriobin.com/post/U5kFRudO

3 sec google search

*looks through list*

That's... a lot of...

*sees 128 to 128*

WHAT THE @#!$ IS THAT?! O.O

Either way, thank. :)

[Hurkyl]

The main thing priority did was remove the most common need for belt balancers.

In the old days, the most common usage of belt balancing was not "I want goods spread evenly between several belts" it was "I have goods spread between several belts and I want to pull some off to send in another direction".

The specific problem that tends to arise is that you wouldn't want to fix each junction to a specific belt from the group because you'd have to pay attention to the usage of each individual belt in the group. So you would include belt balancers as necessary to redistribute the goods between the belts and then have your junctions just pull from whichever belt is most convenient.

With the inclusion of priority, a much simpler system is available -- you simply use priority to shove all the goods as far as you can to one edge of the group of belts, and then have your junctions always pull from that edge.

[Zaflis]

While true mainbus has no use for belt balancers, they still do need lane balancers here and there if stuff gets heavily on 1 side of a belt.

But belt balancers still have use for ensuring train wagons get filled and emptied at equal rate.

[RiO]

Well curated list, ready to go:

https://github.com/raynquist/balancer

[knighttemplar1960]

Originally posted by Zaflis:

While true mainbus has no use for belt balancers, they still do need lane balancers here and there if stuff gets heavily on 1 side of a belt.

But belt balancers still have use for ensuring train wagons get filled and emptied at equal rate.

You only need lane balancers if your design uses disproportion draw from a particular lane. you can fix that with design. In a mega base UPS is the resource you run out of first. Splitters draw all most as much UPS as inserters. You should eliminate any redundant splitters where ever possible.

[RiO]

Originally posted by knighttemplar1960:

Originally posted by Zaflis:

While true mainbus has no use for belt balancers, they still do need lane balancers here and there if stuff gets heavily on 1 side of a belt.

But belt balancers still have use for ensuring train wagons get filled and emptied at equal rate.

You only need lane balancers if your design uses disproportion draw from a particular lane. you can fix that with design.

No, you actually can't. Not realistically - anyway.
Inserters favor grabbing from the near lane of a belt, but will still grab from the far lane if that's not possible. The result is that the very rhythm of the recipes completing in the assemblers straddling the belt and the rhythm with which inserters pick up more ingredient from the belt and leave gaps, dictate what inserters further down that belt will do: near, far, or mix of both.

Just mirroring your builds perfectly on either side of a belt won't cut the mustard either. Because then a stack inserter could, when it swings, still 'overfetch' from its near-lane - the near-lane not being full enough anymore -- and hit the far-lane. This would only work if both machines straddling the belt at the same advance along the belt, have their recipes exactly synchronized, to the tick, such that both mirrored inserters will simultaneously first exert their. near lanes and then won't have anything to grab from each other's far lanes. Otherwise, there will be inconsistencies. And they need to remain eternally synchronized; or those inconsistencies will compound into more steadily uneven draw.

It's pretty much intractable to solve, in any other way than inbound lane balancers.
(Which is not that UPS heavy. It's only two splitters for a 1:1 lane balancer.)

[bigulator]

Originally posted by Sagittarius_A*:

https://factoriobin.com/post/U5kFRudO

3 sec google search

Is there a list of balencers like this but only for yellow belts?

[kremlin]

It's important to remember that the balancer isn't going to give you more material out of nowhere, and you are going to be hard limited on your throughput when using balancers. They're a patch for poor factory design where lines are not being fed consistently, when your factory should be built to run consistently.

[Chindraba]

Some things just need balancers. If you wish to fully drain a mine using belts, or mutli-wagon trains, you either have to set up a balancer or expect random idle states. It doesn't require a full-fledged, fractal-based design with unlimited throughput, but it needs some rudimentary effort.

Feeding the mall, where consumption is anything but stable, is another case where balancers are quite handy, though less 'needed' than the mines. Using a bus to feed the mall helps, but that's really just hiding the balancer over several tiles rather than placing it as a discrete unit.

Most everything else, with attention to detail, can be handled without balancers. Even when not using buffer chests at train stations.

[Hurkyl]

Originally posted by kremlin:

when your factory should be built to run consistently.

Not really: demand for factory outputs is inconsistent. Demand for construction materials depends on what you're constructing. Demand for science packs depends on what you're researching. You can smooth things out somewhat with buffers, but you're still not going to get near-exact ratios unless you're letting your factory play you. Except maybe for super post-game when construction requires a truly negligible amount of resources and you've promised to research the infinite techs in a fixed pattern (and keep expanding the milsci buffers as necessary)... but by then you're not using belts for bulk transport.

You can design for dedicated resource feeds that would halt when factory components go idle. Or you can design a self-balancing system where you let excess resources overflow into other factory components. Both ways make sense; I personally prefer the latter.

[kremlin]

Originally posted by Hurkyl:

Originally posted by kremlin:

when your factory should be built to run consistently.

Not really: demand for factory outputs is inconsistent. Demand for construction materials depends on what you're constructing. Demand for science packs depends on what you're researching. You can smooth things out somewhat with buffers, but you're still not going to get near-exact ratios unless you're letting your factory play you. Except maybe for super post-game when construction requires a truly negligible amount of resources and you've promised to research the infinite techs in a fixed pattern (and keep expanding the milsci buffers as necessary)... but by then you're not using belts for bulk transport.

You can design for dedicated resource feeds that would halt when factory components go idle. Or you can design a self-balancing system where you let excess resources overflow into other factory components. Both ways make sense; I personally prefer the latter.

The demand for science packs is steady, they are all used in equal amounts regardless of what you're researching. That's the end run of your production and is where you most want to fully feed your factory. It shouldn't be treated like something intermittent like a mall.

[Hurkyl]

Which packs are required changes with the technologies though. Even when you've finished the tech tree and doing infinite researches, some require military science and some don't. Some require purple science and some don't.

Unless, I suppose, you've decided to limit yourself just to robot speed and mining productivity. Or some other combination of techs with equal distributions. (which, technically, is the sort of thing I mean by researching them in a fixed pattern; it's just a degenerate example)

[Chindraba]

I look forward to the increased variety of infinite tech options promised for at least Space Age and I think 2.0 base. As it is now they all need Space science and 5 of the other 6. Robot follower count is the only infinite tech to use both Mill and Prod sciences, which probably explains why it's linked to an achievement. All the rest are either production-based and need the grape juice, or military and require the ink.

FFF #376 highlighted the RCU when discussing other infinite researches, which is unfortunate as they're now RIP'd. Still, it needed no Utility or Military science but did need Space science. Somewhere in Reddit Kovarex was discussing that FFF and it was disclosed that there at least one productivity research, steel if memory serves, will be available pre-space, and I think it was even pre-blue. Still, more variety in the resource sink will be welcome and can keep even the end-game mega-base phase interesting.