Splitters

[Santos]

Just wondering if it was possible to add in a future update the ability to edit splitters? (the amount of items pass through before the next item is split off onto another belt)

i.e. for every 3 plates I want 2 to split off in 1 direction and allow 1 plate passes through?

Or maybe this has been already modded in?

[Grandaddypurple]

I do secretly hope for single splitter balancing option but I doubt it'll be implemented.
Currently you can achieve such balancing using combos of splitters/mergers like this:
https://i.redd.it/tx9gt3whtfm31.png

[Fenix]

There is Probably Mod, but your example could be done with a 1 Splitter and 1 Merger.
Run 2 output ports of a splitter to a merger, will give you a 2/1 split.

[Santos]

It ends up taking up a lot of room though where as the option to simply split various amounts each way from 1 splitter would be a lot nicer and seems like the smarter thing to add to the game.

[The Big Brzezinski]

Smart splitters offer most of this functionality as well. You can force a feed to fill up a smaller belt, then manipulate that feed and/or the remainder however you require. It can be very handy when you need to enforce rationality on an unruly production line.

[McCloud]

Would remove the foundation of the game if they did that, if you look at it analytically; it’s practically the core of the game to find ways to overcome the mathematical problem with logistics.

[Grandaddypurple]

Originally posted by The Big Brzezinski:

Smart splitters offer most of this functionality as well. You can force a feed to fill up a smaller belt, then manipulate that feed and/or the remainder however you require. It can be very handy when you need to enforce rationality on an unruly production line.

That doesn't always work, unless your main belt is full (or almost full and the belt you're trying to force feed has a low throughput)

[The Big Brzezinski]

Originally posted by Grandaddypurple:

Originally posted by The Big Brzezinski:

Smart splitters offer most of this functionality as well. You can force a feed to fill up a smaller belt, then manipulate that feed and/or the remainder however you require. It can be very handy when you need to enforce rationality on an unruly production line.

That doesn't always work, unless your main belt is full (or almost full and the belt you're trying to force feed has a low throughput)

It works anytime you can draw away an entire belt's worth of throughput. It doesn't matter how full the input flow is, so long as it fills the measured belt's capacity and the overflow belt is capable of handling the remainder.

And if you are dealing with less than 60 parts per minute, there are much simpler ways of distributing it. I'd probably divide such small amounts right as they come out of the machines, then recombine them into whatever flows I needed elsewhere. It's often better to leave the outputs from one step separated if combining them is just going to make things harder later.

[Man's Best Friend]

Originally posted by Grandaddypurple:

I do secretly hope for single splitter balancing option but I doubt it'll be implemented.
Currently you can achieve such balancing using combos of splitters/mergers like this:
https://i.redd.it/tx9gt3whtfm31.png

What amuses me about the fifths setup is that there's literally no reason to use the 20% or 40% versions when the 80% and 60% versions respectively will give you exactly the same results, but with fewer buildings and belts.

I mean, unless you really really really really need to have that one line go perfectly straight.

[Grandaddypurple]

Originally posted by The Big Brzezinski:

It works anytime you can draw away an entire belt's worth of throughput. It doesn't matter how full the input flow is, so long as it fills the measured belt's capacity and the overflow belt is capable of handling the remainder.

Not anytime. It won't break, but it'll end up working just like an overflow system, I've covered that in another thread I'll just quote it here

Originally posted by Grandaddypurple:

Now let me precisely break down what happens on a belt if you take a single constructor making Cast Screw recipe overclocked to 150%, making screws at 75/min:

20 screws will come out of the constructor on an MK2 belt. The belt reaches the smart splitter with the MK1 bottleneck : the first screw goes onto the belt as intended, and the next screw comes 0,5 seconds later: it's too fast for your mk1 belt that can only handle one item per second. So the second screw goes into the smart splitter's internal storage. A couple of screws make it onto your MK1 belt until the internal storage is full, and there's still 15 screws left to sort : they will now be spread evenly between your "60/min" belt and your "15/min overflow" belt (because this is what's happening, it's overflowing). during this time, your "60/min" belt will have the throughput intended (60/min because it's full), but the throughput of your "15/min" belt is through the roof.
Now comes the second phase : the empty phase
During this time, your constructor took a whooping 16 SECONDS to make the second batch of 20 screws. During those 16 seconds, your MK2 belt is miserably empty. Where will your 60/min belt take its items from? It's supposed to be full, there should be a constant flow of items going into it. There is not. Your 60/min belt is not a 60/min belt. It is a 45/min belt, or 50 at best. And your 15/min is a 25/min belt. It will back up until its own belt is full and the items will finally be able to overflow into the 60/min belt. You realize your smart splitter is not smart at all despite the AI limiter you put inside for which you allocated way too much copper for your taste.
It is a sorting asset, not a balancing asset.

[The Big Brzezinski]

Originally posted by Grandaddypurple:

Not anytime. It won't break, but it'll end up working just like an overflow system, I've covered that in another thread I'll just quote it here

Originally posted by Grandaddypurple:

Now let me precisely break down what happens on a belt if you take a single constructor making Cast Screw recipe overclocked to 150%, making screws at 75/min:

20 screws will come out of the constructor on an MK2 belt. The belt reaches the smart splitter with the MK1 bottleneck : the first screw goes onto the belt as intended, and the next screw comes 0,5 seconds later: it's too fast for your mk1 belt that can only handle one item per second. So the second screw goes into the smart splitter's internal storage. A couple of screws make it onto your MK1 belt until the internal storage is full, and there's still 15 screws left to sort : they will now be spread evenly between your "60/min" belt and your "15/min overflow" belt (because this is what's happening, it's overflowing). during this time, your "60/min" belt will have the throughput intended (60/min because it's full), but the throughput of your "15/min" belt is through the roof.
Now comes the second phase : the empty phase
During this time, your constructor took a whooping 16 SECONDS to make the second batch of 20 screws. During those 16 seconds, your MK2 belt is miserably empty. Where will your 60/min belt take its items from? It's supposed to be full, there should be a constant flow of items going into it. There is not. Your 60/min belt is not a 60/min belt. It is a 45/min belt, or 50 at best. And your 15/min is a 25/min belt. It will back up until its own belt is full and the items will finally be able to overflow into the 60/min belt. You realize your smart splitter is not smart at all despite the AI limiter you put inside for which you allocated way too much copper for your taste.
It is a sorting asset, not a balancing asset.

What you're describing is not a smart splitter failing to divide a 75 screw/minute flow. It's a description of what happens when you fail to deliver a 75 screw/minute flow. What you're actually delivering is a flow of 120 screw/minute for ten seconds, followed by 0 screws/minute for six seconds. This is why I said to portion smaller volumes using different sorting methods. It's like I said:

Originally posted by The Big Brzezinski:

It works anytime you can draw away an entire belt's worth of throughput. It doesn't matter how full the input flow is, so long as it fills the measured belt's capacity and the overflow belt is capable of handling the remainder.

And if you are dealing with less than 60 parts per minute, there are much simpler ways of distributing it. I'd probably divide such small amounts right as they come out of the machines, then recombine them into whatever flows I needed elsewhere. It's often better to leave the outputs from one step separated if combining them is just going to make things harder later.

If you're not sending a consistent enough flow down the belt, a smart splitter isn't the right tool for this job. That's not a fault of the splitter. That's a design mistake.

Again, the idea is that you're subtracting one belt speed's worth of parts away from a flow, leaving the rest as overflow. If the subtrahend you're sending down the first belt leaves no remainder to overflow down the second, it means the minuend you fed into it was too small.

[Grandaddypurple]

Originally posted by The Big Brzezinski:

^

Uuh, there are a thousand ways a flow of 75items/minute can materialize and a flow of 120/min for X seconds and a flow of 0/min for Y seconds is definitly one of them. You're calling it that "not 75/min" just for the sake of argument. It literally comes out of a constructor saying 75/min and will make any machine needing 75/min input work at 100% efficiency.
Really, rule of thumb, if your main belt does not carry its max throughput, don't use the smart splitter to extract a lower tier belt's worth of throughput. Sometimes it will work, sometimes it'll just work the same as if you put a normal splitter, sometimes it will straight up make your system not work properly

[The Big Brzezinski]

In this case, it's happening specifically because you're overclocking a single cast screw constructor to 150%. It's like an engine with just one piston. It's useful for things, but you have to take into account that it's only delivering power intermittently. Understanding what the mechanism is actually doing is the key to keeping it running smoothly.

What a subtracting smart splitter needs is consistency. You can achieve this by saturating the belt capacity. Multiple machines running the recipe at offset times helps, too. Using the minimum tier belt possible helps spread out loads works as well.

Whatever mechanism you use to regulate the belt, the good news is that this regulation is durable. It hardly matters how you divide or subtract from a belt if the peaks and valleys in its flow has been flattened out earlier. You pull 30 ppm from a consistent 120 flow on a mk2 belt, you now have a consistent 30 and 90 flows. You can keep slicing up that 90 flow with smart splitters until it drops below 60.

[Grandaddypurple]

You're desperately trying say my example isn't valid by twisting entire game mechanics lol, I can find many other similar scenarios. And I'm curious to know how you would "flatten" my cast screw 75/min belt
My point was this

Originally posted by The Big Brzezinski:

It works anytime you can draw away an entire belt's worth of throughput. It doesn't matter how full the input flow is

isn't true. No matter how much you twist game mechanics and scenarios