What is wrong with this setup?

[TBog]

I expect to have the same amount of resources in each wagon. What I get is 887 in the first wagon (as shown in the picture) and 1.3k in each of the other wagons.
Is the splitter not merging equally or is my design using them wrong?

[Harkonnen604]

I'm not belt guru, but from my games I saw many times that splitters behave oddly when output is clogged. Anyway, situation in your screenshot looks impossible if output chests were empty by the time train arrived just because they hold more than a wagon, no matter how belt line is organized. Are you 100% sure chests were full to same capacity by the time train stopped?

[TBog]

Are you 100% sure chests were full to same capacity by the time train stopped?

I made sure the train and the chests were full before testing. Over time the difference grows. It's not a big issue, just wanted to make sure I was not doing smth stupid.

[Frightning]

You appear to have 3 saturated regular belts feeding into 1 fast belt. A fast belt can only handle twice the throughput of a regular belt, you either need a 2nd fast belt or upgrade the existing fast belts to express (this is more resource intensive, but simpler to implement in that you need not change the layout).

[TBog]

I will try to upgrade to express belt and splitter and see if it's resolved.
Still I would expect to have the same amount of items taken from each regular belt.

[Frightning]

I will try to upgrade to express belt and splitter and see if it's resolved.
Still I would expect to have the same amount of items taken from each regular belt.

The answer as to why that isn't what's happening can be found in the first 3 splitters and how the inputs are connected to them (and has to do w/ how splitter's behave). You have 2/3 belts going into the easternmost splitter, which effectively are being into a single saturated fast best on it's output side, immediately that is fed into the topmost of the 3 splitters. The remaining belt is fed into it's own splitter, also receives the loop from the other side of the topmost splitter. What happens is that the topmost splitter will take from input evenly, and split it onto the two outputs, this shoves half of total throughput into the loop back into the left splitter, whereas the other half goes on through. This means that the leftmost splitter now has 1/2+1/3=5/6 of the total input coming at it, but that splitter's throughput max is only 2/3 of total throughput, sit it bottlenecks and backlogs. The easiest way to solve this is by getting rid of the loop and making the topmost splitter an express splitter. If you don't have Express Belts handy, you can also achieve similar results by making the loop out of basic belts instead of fast belts, this will ensure that no more than 1/3 of the input can end up on the loop, and hence no more than 2/3 total can end up hitting the leftmost splitter.

As a side note, there is a much simpler way to make your lane balancer than what you have, just put a single splitter down as you have at the start, and then take each output and side-load them onto the desired output belt (make sure the outputs of the splitter enter the single output belt in a T-junction as the 'side-street' of the junction).

[zebediah49]

I agree that it's due to the fast belt on the load balancer, and that it can be fixed by replacing it with a normal belt, but disagree with the numbers --

Let's try to take a single item out of the output, assuming the system is fully backed up (which appears to be your use case) -- there is a 1/2 chance for each of the two lower splitters. For the eastern, that's 1/4 to come from each of the upper two inputs. For the western, however, the combination of fast and regular belt means that there's a 1/6 (1/3 times 1/2) chance to come from the third inputs, and a 1/3 (2/3 times 1/2) chance for it to come from the "retry" belt. We can try again if it comes from retry -- and when we normalize that (so that in the end it always comes from an input), we get 3/8ths from the top two and 1/4 from the bottom one. That's a 3/2 ratio of the drain, which pretty much exactly what you see in the drain from the cars (2000-900 = 1100 to 2000-1300= 700).

[Frightning]

I agree that it's due to the fast belt on the load balancer, and that it can be fixed by replacing it with a normal belt, but disagree with the numbers --

Let's try to take a single item out of the output, assuming the system is fully backed up (which appears to be your use case) -- there is a 1/2 chance for each of the two lower splitters. For the eastern, that's 1/4 to come from each of the upper two inputs. For the western, however, the combination of fast and regular belt means that there's a 1/6 (1/3 times 1/2) chance to come from the third inputs, and a 1/3 (2/3 times 1/2) chance for it to come from the "retry" belt. We can try again if it comes from retry -- and when we normalize that (so that in the end it always comes from an input), we get 3/8ths from the top two and 1/4 from the bottom one. That's a 3/2 ratio of the drain, which pretty much exactly what you see in the drain from the cars (2000-900 = 1100 to 2000-1300= 700).

The loop itself is all fast belt I believe. Also a piece of ore might go through the retry loop many times before it eventually makes it onto the exit belt (so your proposed probabilities are not accurate; it need not have a uniform probability of being sourced from any of the three sets of chests). Unless 0.13 changed it, cargo wagons hold a max of 1500 ore, not 2k, which gives about 200 missing from upper wagons and 613 from bottom wagon, given that it would take some time for saturation of belts to occur, and that there is more ore on the belt for the bottom wagon than the other two. I propose that the actual ratio is roughly 1/2 from bottom wagon to 1/4 from each or top wagons (long term trend here).

[brunzenstein]

or use madzuri's combinator managed smart loader and the problem is fixed the easy way
https://youtu.be/eWGXzvEHl9k

[siggboy]

or use madzuri's combinator managed smart loader and the problem is fixed the easy way
https://youtu.be/eWGXzvEHl9k

This is an unloading station, not a loading station...

[kinnom]

or use madzuri's combinator managed smart loader and the problem is fixed the easy way
https://youtu.be/eWGXzvEHl9k

This is an unloading station, not a loading station...

the only difference between loading and unloading is that one takes the items from the belt and puts them in a train and the other takes the items from the train and puts them on a belt.

[siggboy]

or use madzuri's combinator managed smart loader and the problem is fixed the easy way
https://youtu.be/eWGXzvEHl9k

This is an unloading station, not a loading station...

the only difference between loading and unloading is that one takes the items from the belt and puts them in a train and the other takes the items from the train and puts them on a belt.

He suggested to use a smart loader to solve an UNLOADING problem, which will not work.

There is a smart UNloader, but it works differently from the loader.

So the difference is not as simple as you suggest (if you don't know how the smart (un)loader works you should look it up ).

[brunzenstein]

or use madzuri's combinator managed smart loader and the problem is fixed the easy way
https://youtu.be/eWGXzvEHl9k

This is an unloading station, not a loading station...

the only difference between loading and unloading is that one takes the items from the belt and puts them in a train and the other takes the items from the train and puts them on a belt.

He suggested to use a smart loader to solve an UNLOADING problem, which will not work.

There is a smart UNloader, but it works differently from the loader.

So the difference is not as simple as you suggest (if you don't know how the smart (un)loader works you should look it up ).

"siggboy " Your mistaken - It really does matter whether you load or unload based on Mazuris's smart loader as the destination of the well sorted items is always a storage box in-between .

[siggboy]

"siggboy " Your mistaken - It really does matter whether you load or unload based on Mazuris's smart loader as the destination of the well sorted items is always a storage box in-between .

Why am I mistaken? I just said that it DOES matter. The smart loader can not be used for unloading and vice versa.

[brunzenstein]

"siggboy " Your mistaken - It really does matter whether you load or unload based on Mazuris's smart loader as the destination of the well sorted items is always a storage box in-between .

Why am I mistaken? I just said that it DOES matter. The smart loader can not be used for unloading and vice versa.

I mistyped sorry "It does not matter" is correct - ...

[siggboy]

Ok, so you use the smart loader to unload from a wagon to buffer chests.

First of all, why would you do that? The buffer chests get filled evenly when you unload a train (without any combinator help). It makes no sense to use the smart loader for unloading a train.

Second of all, how does it help with OPs problem? The train has already been unloaded, now the buffer chests empty unevenly because of the belt layout.

What he could do, is to use a balanced UNloader, which would empty the buffer chests evenly. This lowers throughput a little, but it works. It is not what you've suggested, however. The balanced unloader requires a different circuit from the balanced loader (I know both, I've used both, and I understand the difference).

[brunzenstein]

Ok, so you use the smart loader to unload from a wagon to buffer chests..

Alternatively use:

[siggboy]

Argh.

[brunzenstein]

Argh.

here is a working gif ®decay42 https://gfycat.com/ForthrightQualifiedHuman
from: https://www.reddit.com/r/factorio/comme ... ading_and/
there you also find a blueprint of the unloader

[siggboy]

Yes, it's a layout that works. I don't understand why you keep posting it, but thanks.

It's not really effective though, since it compresses the output from 12 (!) stack inserters to one single blue belt. You can get the same result with 4 inserters, and then the layout becomes a lot simpler. If you need to create the 3 compressed belts that are possible, you need an entirely different layout.

So in short, for 0.13 this unloading station is craptastic.

The loading has the same problem, splits 1 blue belt into 12 lanes, does not even remotely exploit the throughput of the inserters (and could be much simpler for that reason).

All of this, again, has nothing to do with the question that was asked to start this thread.