Compact tileable 1-belt unload station

[huliosh]

Splitter in your case isn't necessary.
edit:

BTW they don't work synchronously. %)

You need an extra empty space on belts to make it work synced, cause there is a difference in length and you need some space to build up the compressed belt. Also 6 stack inserters are an overkill for any single belt.
(1 blue belt = 1.5 red belt)

6 stack inserters filling 2 red belts

belt compression

belt_compression.jpg (288.37 KiB) Viewed 3312 times

Splitter helps a lot to fill gaps in the lanes from another belt

No it doesn't help to fill gaps. The only thing it does is balance between 2 belts. You gonna get evenly distributed gaps instead. But I agree that it helps to balance though.
This way you gonna fill all the gaps:

belt compression2

belt_compression2.jpg (352.52 KiB) Viewed 3303 times

[LazyLoneLion]

Splitter in your case isn't necessary.

Splitter helps a lot to fill gaps in the lanes from another belt. Belt before splitter works as a small buffer and one belt fills gaps in another belt thanks to the splitter.

Splitter helps to balance lanes on the belt if the consumption is unbalanced too.

[LazyLoneLion]

What might also help is to connect the inserters to the belts after where the inserters drop things. Set the belts to report and hold the item count and the inserters to work when item count == 0. This still relies even more on the faster belts for compacting the gaps away but I found this prevents alternating between one and all others inserters if you connect to the belts in the right distance from the inserters.

BTW this design works perfectly.
You really need "all-blue" station, with few blue belts after the splitter as well. You only need to read contents of one belt below the bottom inserter (more belts don't help). And the inserters become totally synchronized, and unloading becomes totally balanced, and red belt will be full and compressed.

Still, my design without circuitry or blue belts is unbalanced but is also cheaper, simpler and on practice works on the same speed.

[mrvn]

I think the most important part is to have a faster belt at the inserters. This has 2 effects:

1) the obvious compression of gaps where the belt slows down later
2) move items away from the inserters drop point so they can drop faster and pick up items from the chest again sooner

[LazyLoneLion]

I think the most important part is to have a faster belt at the inserters.

Or you might just ignore all that circuitry and balancing and use cheap red belts. And have fully compressed belt of ore and the same train unloading time.

[huliosh]

I think the most important part is to have a faster belt at the inserters.

Or you might just ignore all that circuitry and balancing and use cheap red belts. And have fully compressed belt of ore and the same train unloading time.

Yeah, more space for additional stations!

[mrvn]

I think the most important part is to have a faster belt at the inserters.

Or you might just ignore all that circuitry and balancing and use cheap red belts. And have fully compressed belt of ore and the same train unloading time.

Except you don't have the same unloading time. and we are back at the beginning, going round in circles.

[LazyLoneLion]

Except you don't have the same unloading time. and we are back at the beginning, going round in circles.

I've shown that unloading time is the same. Because it is limited not by free space in the chests, but by free space on the belt. And the belt is either free or backed up. In a case the belt is free -- chests will have free space even without superfast belts and circuitry. In the case the belt is backed up -- a train will be unloaded not sooner than the belt will consume enough ore. Not sooner, be it fast belt with circuits or simple inserters with red belt.

[mrvn]

Except you don't have the same unloading time. and we are back at the beginning, going round in circles.

I've shown that unloading time is the same. Because it is limited not by free space in the chests, but by free space on the belt. And the belt is either free or backed up. In a case the belt is free -- chests will have free space even without superfast belts and circuitry. In the case the belt is backed up -- a train will be unloaded not sooner than the belt will consume enough ore. Not sooner, be it fast belt with circuits or simple inserters with red belt.

I don't know what you think you have shown but clearly when the belt consumes items somewhat slowly only the first chest will empty. Then when the train comes in it takes 6 times as longs to unload as with balanced chests. And this matter if the train has other places to be as well. If the consumption speeds up the first 2, 3, 4 chests gets emptied to various degrees. The effect remains though.

[huliosh]

As you can see, you don't need splitters to compress the belt. There are splitters at the top, that are balancing between belts, but it doesn't help to compress it.
Top is Loading station(B), at the bottom are 2 Unloading stations (A).
TopA has even balance for input and output inserters, BottomA has only even chest loading.
The length of the buffer belt lane is enough to fill the gap between output inserter balance delay.

train_supply.gif (11.03 MiB) Viewed 5663 times

Buffer and balance

buffer and balance.jpg (1.11 MiB) Viewed 5660 times

Correction: Not average count, but just (chest item count) > (average - inserter stack size)

[LazyLoneLion]

I don't know what you think you have shown but clearly when the belt consumes items somewhat slowly only the first chest will empty. Then when the train comes in it takes 6 times as longs to unload as with balanced chests.

It has been said a few times already. I have nothing to do but repeat it, trying to explain again.
If "the belt consumes items somewhat slowly", that means that your trains wait to unload. They do not wait to load, because loading happens at the top speed with all the balancing and often on several outposts simultaneously. One train is empty -- next train comes to unload station instantaneously or almost instantaneously. So, there is no free space in the chests to unload whole train there. A train has to wait for the belt to free some space. So everything depends only on the consumption rate on the belt. Anyway, we have leeway so the belt will be always full if its consumption is low -- and that's all we need in the end.

But the train will wait the same time -- it will wait until the belt will consume one car of the ore. There will be limit on the belt consumption, not on the unload speed.

BTW it will not take "6 times as long" anyway, because we have two equal chests on top, balanced by the splitter, not the only one. So you should have said "3 times as long". And it would be wrong too. Because read above -- it's not the chest free space that is limiting us. It's backed up belt.

Let's imagine we have backed up belt and simple "unbalanced" setup. So all chests are full, including the top one (two in fact - #1 and #4, because they feed different lanes, balanced by the splitter). The top one couldn't be unloaded faster than loaded -- the inserter speed is the same anyway. If consumption rate is less than chest is loaded, then the chest is full. If belt consumption rate is bigger than one chest/inserter unload rate, then two chests will be unloaded successfully and thus loaded from the train to the "full" state again. Train-to-chest rate is always bigger than chest-to-belt, so chests are full. One or two might be "almost full", in the process of simultaneous loading-unloading.

But the train still waits for the belt to consume enough ore. "Enough" equals "one carriage" of ore. Be it unloaded by one chest or by two/three (but every one on proportionally slower rate, because of the balancing and not enough free space on the belt) it will wait for the belt, and the consumption speed of the belt doesn't depend on the unloading.

There is only one advantage in the expensive and complex balanced station -- it will give full red belt until it is empty. Unbalanced will consequently unload top two chests (#1 and #4), then middle two (#2 and #5, partly unloaded at the moment) and then bottom two (#3 and #6, partly unloaded as well) with the belt 80%-full until last chests are empty as well or the new train comes in (at last!).
It's not about a train unloading time, but about the station unloading time when there is fast consumption and no incoming ore.

[mrvn]

I don't know what you think you have shown but clearly when the belt consumes items somewhat slowly only the first chest will empty. Then when the train comes in it takes 6 times as longs to unload as with balanced chests.

It has been said a few times already. I have nothing to do but repeat it, trying to explain again.
If "the belt consumes items somewhat slowly", that means that your trains wait to unload. They do not wait to load, because loading happens at the top speed with all the balancing and often on several outposts simultaneously. One train is empty -- next train comes to unload station instantaneously or almost instantaneously. So, there is no free space in the chests to unload whole train there. A train has to wait for the belt to free some space. So everything depends only on the consumption rate on the belt. Anyway, we have leeway so the belt will be always full if its consumption is low -- and that's all we need in the end.

And that is where you are wrong. A train has to wait for the chests. The chests have to wait for the belt.

The problem is that you are only considering a single purpose train. Your argument is that the chests will empty out at the speed of the belt and that if that is more than one inserter can unload from a train then more chests will become not-full and more inserters will unload. So unloading will speed up as consumption increases. So the system will balance itself (assuming the buffer chests are large enough to reach an equilibrium without ever running empty, try wooden chests .

It is true that the frequency of trains arriving needed is determined by the consumption. All that changes by balancing is the ratio between loading/unloading and doing something else. Not the frequency of visits. A balanced load/unload will have more time in between. And that means the same train can pick up and deliver something else before picking up the next load. If one station starts holding up the train because the buffer chests are unbalanced then less time remains for the other stations if the same frequency is to be kept. Note: the station blocking the train will not be the one running dry, which is where your argument breaks.

[LazyLoneLion]

If one station starts holding up the train because the buffer chests are unbalanced then less time remains for the other stations if the same frequency is to be kept. Note: the station blocking the train will not be the one running dry, which is where your argument breaks.

You didn't get my argument at all.
I've not said that emptied station will block the train. On the contrary -- only backed up station will block the train and the time the train is blocked _is_ determined by the consumption on the belt only. So, backed up station will block the train for the same time if it's balanced or not.

There will be difference in unloading station, as I said. And there can be a small difference on unloading a train if the station and the belt were almost empty and two full trains came in succession.

...you are only considering a single purpose train.

Yes, I'm considering a single purpose train. It has its logic -- as with 1-loco-1car trains.
But can't see the difference if it's multipurpose.

A train has to wait for the chests. The chests have to wait for the belt.

...and that's why the train has to wait for the belt. What I've said. It's obvious.

then more chests will become not-full and more inserters will unload. So unloading will speed up as consumption increases.

It will speed up as consumption increases be it balanced or not. If (!) consumption increases. The bottleneck is the belt, not the quantity of not-full chests. If more chests are unloaded simultaneously, they will be unloaded not as often as fewer chests in an unbalanced design.

try wooden chests

If you look closer on the youtube link, you will see I have used wooden chests. Proves nothing. Anyway plenty of time for a train to come with another load of ore. With wooden chests it takes 70secs to unload a carriage to the red belt (when chests are full, and they stay full afterwards), ~110 seconds after that till the first topmost chest is empty, ~80sec till second is empty and then the belt becomes visibly not compressed enough and it takes just ~15 seconds to empty last chest. Like 3+ minutes of total unload time with a departed train and full wooden chests. With steel chests it will buffer even more and next train will definitely come, so we will have constant full red belt. Without expensive or complex unloading station.

[mrvn]

You are still not getting it.

Say the belt consumes 600 units in the time the train takes for a round trip.

If you unload the chests balanced then every chest will have 100 units space when the train comes by next and unloading will happen with 6 inserters in parallel.
If you unload unbalanced the first chest will have 600 units space and all others will be full. Unloading will happen with 1 inserter and take 6 times as long.

As you can see from this simple example the speed of the belt does not change but unloading time does. The train waits on the chests, the chests wait on the belt. That does not mean the train waits on the belt. The buffering of the chests does exactly that: it buffers. It uncouples the belt and the train.

[torne]

You are still not getting it.

Say the belt consumes 600 units in the time the train takes for a round trip.

If you unload the chests balanced then every chest will have 100 units space when the train comes by next and unloading will happen with 6 inserters in parallel.
If you unload unbalanced the first chest will have 600 units space and all others will be full. Unloading will happen with 1 inserter and take 6 times as long.

As you can see from this simple example the speed of the belt does not change but unloading time does. The train waits on the chests, the chests wait on the belt. That does not mean the train waits on the belt. The buffering of the chests does exactly that: it buffers. It uncouples the belt and the train.

I agree that it does make a difference, but this example isn't a great one: if the consumption rate is low enough that there won't be room in the chests to fully unload the train, then once the chests fill up the train will have to wait on the consumption anyway, and it makes little difference how many inserters you have unloading the train.

There's two cases I can think of where it makes a demonstrable difference whether you have balanced buffer chests or not, and both of them are only relevant if when the train arrives, there *is* enough room to unload the entire train without waiting for more resources to be consumed:

1) If your station is used to unload multiple different things, or if the location where other, queueing trains wait blocks other tracks. In this case, you need to unload every train as fast as possible even if it makes no actual difference to the rate at which resources flow off down the belt, because you want this train out of the way ASAP to unblock the rest of the train system. You can avoid this by just not building a train system that has these problems, though.

2) If the rate of consumption of the resource is faster than the rate at which your trains can deliver it. In this case you also want to unload every train as fast as possible, because the belt will *never* back up, and so any time you save by unloading slightly quicker is time that the train can use to start going back to where the resource comes from and loading itself back up again, increasing the total throughput.

So, as far as I can think of (there may be other circumstances that haven't occurred to me, so this is definitely not conclusive), balanced buffer chests aren't important as long as your station design doesn't cause waiting trains to block other unrelated trains (ones delivering a different resource), and your trains are delivering enough resources to keep up with or exceed your consumption rate. If one of those isn't true, then you'll get more throughput from your system by adding a balanced buffer.

If there's anything I'm overlooking or mistaking here, please let me know, but as far as I can see, you're both half-right

[mrvn]

You are still not getting it.

Say the belt consumes 600 units in the time the train takes for a round trip.

If you unload the chests balanced then every chest will have 100 units space when the train comes by next and unloading will happen with 6 inserters in parallel.
If you unload unbalanced the first chest will have 600 units space and all others will be full. Unloading will happen with 1 inserter and take 6 times as long.

As you can see from this simple example the speed of the belt does not change but unloading time does. The train waits on the chests, the chests wait on the belt. That does not mean the train waits on the belt. The buffering of the chests does exactly that: it buffers. It uncouples the belt and the train.

I agree that it does make a difference, but this example isn't a great one: if the consumption rate is low enough that there won't be room in the chests to fully unload the train, then once the chests fill up the train will have to wait on the consumption anyway, and it makes little difference how many inserters you have unloading the train.

There's two cases I can think of where it makes a demonstrable difference whether you have balanced buffer chests or not, and both of them are only relevant if when the train arrives, there *is* enough room to unload the entire train without waiting for more resources to be consumed:

1) If your station is used to unload multiple different things, or if the location where other, queueing trains wait blocks other tracks. In this case, you need to unload every train as fast as possible even if it makes no actual difference to the rate at which resources flow off down the belt, because you want this train out of the way ASAP to unblock the rest of the train system. You can avoid this by just not building a train system that has these problems, though.

2) If the rate of consumption of the resource is faster than the rate at which your trains can deliver it. In this case you also want to unload every train as fast as possible, because the belt will *never* back up, and so any time you save by unloading slightly quicker is time that the train can use to start going back to where the resource comes from and loading itself back up again, increasing the total throughput.

So, as far as I can think of (there may be other circumstances that haven't occurred to me, so this is definitely not conclusive), balanced buffer chests aren't important as long as your station design doesn't cause waiting trains to block other unrelated trains (ones delivering a different resource), and your trains are delivering enough resources to keep up with or exceed your consumption rate. If one of those isn't true, then you'll get more throughput from your system by adding a balanced buffer.

If there's anything I'm overlooking or mistaking here, please let me know, but as far as I can see, you're both half-right

I totally agree with 1. That's the case I'm talking about. It's the time spend in the station that matters, not the amount unloaded on average (as that is limited by the belt). And yes, it's something you can avoid by designing your trains differently. But it's also very tempting, especially at the start where adding a new train just so it can wait somewhere while other trains are only 50% used is a hassle. Mostly I use this with coal. Have it drop of part of it's load at every station. At some point you have the train set so it can just about keep up with demand and then balanced or unbalanced unloading means the world.

As for 2 I think that is a non issue. If the train can't deliver goods fast enough then all chests will empty out. So weather you empty them out balanced or not doesn't matter to the train. When it arrives all chests are empty and unloading will be as fast as possible. Only thing you can do is use more chest (e.g. double sided unloading for twice the speed) or use more trains so the next one is waiting to unload when the first one finishes. Unless the problem is you are producing/loading too slow. Then fix that end, e.g. build more iron ore mines so the train doesn't have to wait to be filled.

Also there is a big window between running dry and not having enough space in the buffer chests to fully unload. With unbalanced unloading it will find some equilibrium where the number of chests with space for unloading produces the right unloading speed. That's what you usually end up with,

[torne]

As for 2 I think that is a non issue. If the train can't deliver goods fast enough then all chests will empty out. So weather you empty them out balanced or not doesn't matter to the train. When it arrives all chests are empty and unloading will be as fast as possible. Only thing you can do is use more chest (e.g. double sided unloading for twice the speed) or use more trains so the next one is waiting to unload when the first one finishes. Unless the problem is you are producing/loading too slow. Then fix that end, e.g. build more iron ore mines so the train doesn't have to wait to be filled.

Here I'm comparing to not having buffer chests *at all*, not just to having no balanced unloading logic. If you aren't blocking the rest of the train system and you're producing more than you're consuming, you don't need any chests at all and unloading directly to the belt will work fine.

I wasn't really considering the case of having a buffer but letting it unload sequentially, which I now realise is what the thread started off talking about.

So.. I think I do actually agree that there doesn't seem to be a case where an unbalanced buffer is useful. If you're in a situation where it not being balanced doesn't matter (which does exist), then not having a buffer at all also wouldn't matter, and you could save even more space by just unloading to the belt. I think?

[mrvn]

Unloading from the train directly has merits for coal for example. Coal usage is often slow enough that the train can unload, drive away to get more coal and return before the little buffering the belt provides has run out. I have several stations now where I only have a single wooden chest as coal buffer. I didn't go quite so far as having none.

Which makes me think. Who has used cars as buffer chest? Since they are 2x2 tiles big some balancing would be implicit.

[Nich]

Or just use trains as chests. Have 2 or however many unloading stations you need and trains just sit and wait to unload straight to belt. Not sure how you would prioritize the more empty train or if you even need to. I am thinking of the system bottle neck is past the trains

[mrvn]

Or just use trains as chests. Have 2 or however many unloading stations you need and trains just sit and wait to unload straight to belt. Not sure how you would prioritize the more empty train or if you even need to. I am thinking of the system bottle neck is past the trains

Some people have trains setup so they wait at the station till the next train arrives. I used that for crude oil barrels since waiting for a train full of barrels takes way to long and waiting for a fixed amount would be fine at the start and then cause longer and longer waits as the pump jacks run down. Drawback is that you need at least one extra train.