6:1 and 1:6 blue belt balancers for train stations

[Premu]

I was experimenting a lot with different designs for loading and unloading from trains to belts. I use a side of a train to fill up a single blue belt. My requirements for the balancers is:

- They should fill up/draw from each chest equally
- They should lead to a fully compressed belt (for unloading only)
- They should be compact, i.e. not be wider than six lanes so I can easily place more of those side by side for trains with more carriages.

First I had quite complicated designs which still could lead to not fully compressed belts. While making balancer for 2^n belts is pretty easy, other number of lanes are far more complicated.

Well, until I had finally the right idea for the 6:1 balancer - why not combine a 4 lane and a 2 lane balancer - and use the different belt speeds of yellow, red and blue belts to achieve a balanced pull? While in general this is not a good idea to rely on - it works with fully filled belts, but not if the input is not compact - it's not a real issue at least for unloading from buffer chests. Here I can ensur that the input is either full or nothing comes at all once the buffer chests run empty.

You can't inverse this for the loading variant, though - here I can't be sure that only a full belt arrives. So here at least an additional loop is necessary.

Still, these two designs (loading on the left, unloading on the right) is the result of my thoughts:

[coppercoil]

You can take 3:1 balancer for unload. Here are lots of balancers.
You can also use 4 chests for every side. Would it be too easy?

[Premu]

Using six chests means that I need a train unloading or loading at the station only 50% of the time to get the actual throughput to fill a blue belt. With only four belts you need to achieve a ratio of 75%. That makes train schedules more robust.

Also, most 1:3 balancers are wider than 3 squares. Those which are just three squares wide tend to be pretty long. So I was searching for something more compact.

[5thHorseman]

Way back in 0.15 I made these two stations and used them for a long time. They're as compact as I could make a balanced setup be. There's plenty of room to fit a balancer or a 3:1 compressor/expander between the stations and the edge of the tracks. I'm not 100% sure this works in 0.18 but I don't think there were any changes to how inserters placed on belts between versions.



These days I actually prefer the simpler, more compact and unbalanced designs, though I've been considering revisiting these now that I have more experience in the game.

[5thHorseman]

Way back in 0.15 I made these two stations and used them for a long time.

I had vague memories that they didn't actually balance well and ... yeah they don't balance well. Backups in any line (caused by - say - inserters not keeping up with the flow) causes imbalance. I think it actually may be more imbalanced than the simpler setups with 2 or 3 inserters per line.

So I made this one. It has wires but no combinators. It's not perfect but it DOES ensure even loading. It just connects all belts to all inserters, reads the belt contents as a hold, and won't let the inserters run unless all the belts are full. This ensures the inserters grab the same number of items each.

While testing I tried not having the straight belt segments after the splitters (and instead counting straight segments before the splitters), but as I expected inserters taking from splitters was inconsistent. It even seemed to load slower than belts but that may have just been observer bias.



EDIT: This one's a bit borked too. I know it's not super important but it still drives me nuts how hard this is. No wonder I gave up years ago and just went with the easy - and lopsided - way out.

[Impatient]

... why not combine a 4 lane and a 2 lane balancer - and use the different belt speeds of yellow, red and blue belts to achieve a balanced pull? ...

That caught my eye.

The reason why not to use different belt speeds to achieve ratio splitting or ratio pulling is because it only works if the input/ouput belts are compressed and run at max speed. If these conditions are not met, there will be no split or pull in the desired ratios.

Your solution showcases that as well. The rightmost blue balancer draws equally from its upper and its lower input, correct? Now the important question: What chests will be empty first, when the items are consumed by the factory at a rate of just 30 items/s ?

[Premu]

... why not combine a 4 lane and a 2 lane balancer - and use the different belt speeds of yellow, red and blue belts to achieve a balanced pull? ...

That caught my eye.

The reason why not to use different belt speeds to achieve ratio splitting or ratio pulling is because it only works if the input/ouput belts are compressed and run at max speed. If these conditions are not met, there will be no split or pull in the desired ratios.

Your solution showcases that as well. The rightmost blue balancer draws equally from its upper and its lower input, correct? Now the important question: What chests will be empty first, when the items are consumed by the factory at a rate of just 30 items/s ?

I know - I wouldn't use that design as general 6:1-balancer for this specific reason. Here I can assure constant supply, though.

What happens if the demand is lower than the maximum? For that I just measuered it in a "real-life" example:



Here you see an unloading station in which two blue belts full of plastic and one blue belt of batteries are unloaded. (Don't mind the belt on the top - I happen to need 2.07 blue belts of it, so I added the top one as light weight solution...)

The plastic is used up almost completely. The batteries on the other hand are needed in far lower quantities, and a blue belt is actually an overkill. I used counters to measure the throughput. A is the plastic on the "yellow" side, B the plastic on the "red" sid, C the batteries on the "yellow" side, and D the batteries on the "red" side. (Only one inserter each)

As you can see, A and B are almost equal. A is only slightly higher than B. So for high demand, the equal ratio is almost achieved. For the batteries it's not the case - C is almost double the value of D, so here it's closer to a 2/1 ratio which is actually to be expected.

Now, is this a real problem? Actually not in my specific usecase. My goal is to ensure high throughput if needed. If the crates are not emptied in an equal ratio, the potential throughput will be lower - but if that's caused by lack of demand, that's a non-issue.

So - this design is definetly not a good general 6:1 balancer, but it works pretty well for unloading trains.

[disentius]

Your have no safeguard for unbalanced lane consumption in your design. Do you have that downstream somewhere?
And it is big. here are two other solutions:
1. belt balanced unload, assuming even lane consumption:

Balanced unload 1.png (545.18 KiB) Viewed 35519 times

2. Balancing chests with a reverse Madzuri, between wagon and unload chests.

inverse madzuri balanced unload.png (508.44 KiB) Viewed 35519 times

[Premu]

Your have no safeguard for unbalanced lane consumption in your design. Do you have that downstream somewhere?

Each crate's content will end up 50% on the left and 50% on the right lane. That is ensured by the three splitters right after the inserters. So even if only one lane is pulled away it will pull from all six crates.

And it is big. here are two other solutions:
1. belt balanced unload, assuming even lane consumption:

That's not a 6:1 balancer using one side only, though, but an 8:1 balancer using both sides of the train. I have to say the idea using the fish to prevent stuff clogging up in the dead ends of the splitters is great, though.

2. Balancing chests with a reverse Madzuri, between wagon and unload chests.

I don't see how this is in any way balanced. The two inserters in the middle won't be able to unload anything at all until the crates on the edges run empty. Or do I oversee something here?

Something like this was actually my very first design for an unloading station, and I ran into exactly this problem.

[disentius]

1. You are right, i overlooked the balancers

2. You mentioned using 6 chests to reduce train traffic. This solution makes sure chests are constantly "rebalanced" when unloading a train. They then become unbalanced while unloading, but as long as you have a train there before the outer chests are empty, it works.

[kjagodka]

Those are my most compact designs for:
-unloader which works only for fully consumed belt
-unloader which works even with unevenly consumed belt
-loader which works only for compressed belt
-loader which works even with uncompressed belt
but they are all so simple, I'm probably not the first one to come up with such designs.

I know that some of blue belts in those designs could be replaced by red/yellow belts, but this way it works with lower inserter's stack capacity because inserters wait less for the item to move underneath on the belt.
For example this unloader design works with inserter's stack size of 4, but it would require stack size of 5 if inserters would place items on red belts.

[Premu]

Those are my most compact designs for:
-unloader which works only for fully consumed belt
-unloader which works even with unevenly consumed belt
-loader which works only for compressed belt
-loader which works even with uncompressed belt
but they are all so simple, I'm probably not the first one to come up with such designs.

I know that some of blue belts in those designs could be replaced by red/yellow belts, but this way it works with lower inserter's stack capacity because inserters wait less for the item to move underneath on the belt.
For example this unloader design works with inserter's stack size of 4, but it would require stack size of 5 if inserters would place items on red belts.

These are some nice designs. The first one is pretty similar to mine, you just used the fact that you can use a underground belt to prevent the belt forming a curve. This saves a lenth of two.

For the second design - the first row of splitters is useless. Inserters will put down the items on the other end of the splitter. That's the reason my design starts with a bunch of normal belts in front of the splitters.

Here's a small example where you can see it: