Motion of the Chest Wave

[PyroFire]

When i speak of the "Chest Wave", i am of course talking about how this thing works:


How does it work?
It seems to pick a lane and stick to it until the lane is blocked or full.

I'm trying to consider ways to create an "overflow valve" where resources will travel down a secondary path ONLY if the primary path is full.

I'm sure this could be achieved using some combinator logic, but would be nice to see if its possible using the built-in logic and math it out.

And if it changes directions along the path, it seems to change which inserter it will use to output as well.

[DOSorDIE]

1 input and 3 output ... really?
So its useless.
But the concept is nice.

Buffer.gif (6.2 MiB) Viewed 4496 times

[Deadly-Bagel]

DOS' buffer is what I came up with too, though I tend to use a single chest and a splitter to merge it back in. Unless it's for green circuits or iron/copper ore/plate, or you''ve got a particularly large factory and want a huge buffer, you generally don't need more as long as your production is solid.

[mergele]

The easiest I could think of would be to make a buffer chest and activate your seond route if the buffer is above a certain threshold.

[ssilk]

How does it work?
It seems to pick a lane and stick to it until the lane is blocked or full.

It would be interesting to find that out an post it, cause I think that could be really useful.

For example: The forum was really fascinated to understand how splitters really work in detail and what else could be done with them.

[Deadly-Bagel]

Man, that topic on splitters was a head spinner.

I don't know how the game picks an inserter to start working first, forum consensus seems to be something to do with the order they are placed in. As far as I've observed once a source is outputting via an inserter, it will prefer that output until that inserter can no longer work (power, removed, destination full).

However I don't know if/how preference changes when:

• more items become available than a single inserter can move
• an item becomes available for output while the preferred inserter is already moving an item/stack
• power to the whole network is lost (including providing source items so as to not trigger either of the above as well)

Would be an interesting experiment, you could probably build some sort of logic processing out of it if you knew all this and had the mind for it.

[ssilk]

"You" in the sense of a) the thread author (PyroFire) b) you all, also called "we" c) you as "myself" d) me (ßilk).

[Deadly-Bagel]

"You" in the sense of whoever is reading it. Personally I have not the time, energy or patience for such things right now ^^

[PyroFire]

Has anyone done throughput comparisons?



This is set up to stop all the final inserters when a total of 500 items have reached the end.
In this first experiment, the chest lane ended with 360 items, and the belt lane ended with 144 items.
This means the chest wave is approximately 250% faster than a blue belt in terms of throughput.

Let's try this with a second inserter for the belt lane, and add 2 more inserters to handle the extra compression.

Chest Lane: 288
Belt Lane: 244.

And now a single inserter chest wave compared to a max compressed blue belt.
The inserters and belts are set like this to create the max compressed belt while having the inserters start at the same location and minimize the loss of compression + time spent travelling around those bends.
If the single inserter has more at the end, the chest wave is faster than even a blue belt.

Running this with the cap at 5000.
Chest Lane: 2100
Belt Lane: 2966

This difference is surprisingly small.
Although this is all done with stack inserter at max level.

However i should also add, in all cases, the chest wave is significantly faster in terms of tiles travelled per second compared to the express belt, so in far longer chains it may end up having higher throughput?
As for how fast, i'll need to do some tests.

[Deadly-Bagel]

If you're testing throughput the ONLY test that had relevance was a compressed blue belt vs the stack inserters, and the belt easily came out on top. Additionally the inserters need to be powered so you also need additional power production which uses more space (it can just be elsewhere) not to mention the staggering amount of extra resources required (EDIT and power lines so you're having to overflow into a second lane)

The only factor the inserters came out on top of was the speed at which the items were transported, therefore you have less items tied up in transit. However as those items are less than the extra resources required to build the inserters, realistically the belts are still more efficient.

Throughput is the same regardless of length, the only difference you'll notice by increasing the distance is the latency. Inserters will deliver the items more quickly initially, but will eventually be overtaken by the belt.

[PyroFire]

If you're testing throughput the ONLY test that had relevance was a compressed blue belt vs the stack inserters, and the belt easily came out on top. Additionally the inserters need to be powered so you also need additional power production which uses more space (it can just be elsewhere) not to mention the staggering amount of extra resources required (EDIT and power lines so you're having to overflow into a second lane)

The only factor the inserters came out on top of was the speed at which the items were transported, therefore you have less items tied up in transit. However as those items are less than the extra resources required to build the inserters, realistically the belts are still more efficient.

Throughput is the same regardless of length, the only difference you'll notice by increasing the distance is the latency. Inserters will deliver the items more quickly initially, but will eventually be overtaken by the belt.

I just ran the very same test again over a distance of 300 tiles with a cap of 5000.


Results;
Chest Lane: 2364
Belt Lane: 2667

By your logic, the chest lane should have ended with less than what it had with the short test.

Maybe there's more to it.

The initial wave through the inserters is *Approximately* 33% faster than the belts.
This wasn't measured, but it looks like it's around there.

And finally, there are situations where a fully compressed blue belt is not required.
Having this information is actually Quite relevant, as it may be useful.
In almost any situation where you know you're not going to receive a fully compressed blue belt, the chest wave will provide far faster throughput.
A good example of a situation where this may be useful is in my tier 3 assembly blueprint, where there isn't a way to easily add a balancer straight into the blueprint.
Instead, having the output of the assemblers flowing through a chest wave will result in significant higher maximum throughput available compared to the previous blueprint, albeit slightly more expensive.
How can you simply say that it's not relevant?

When i finish my new factory, i'll re-build this test at 1000 tiles or longer and use a significantly larger cap, or even just until if or when the chest lane overtakes the belt lane.