Let's see your clever builds

[tobsimon]

Small lane mixer with underground belts:

LaneMixer.png (85.17 KiB) Viewed 13829 times

Without underground belts, I suspect Tallinu's design already is minimal.

[Talguy]

Expandable copper smelting (can be upgraded to electric furnaces later), optimized for beacons and (almost)blue belts. max is 54(missing 2) furnaces, but you can add extra with beacons.

At the bottom is a 4 lane belt balancer.

Also, pre-planning massive production of electronic circuits, based on https://forums.factorio.com/forum/vie ... =8&t=12588. However, I don't really like the double underground belt trick as it requires red belts, I want my factory to be exandable so I used a 1 width tile in between sections (later in the game that will become from top to bottom: yellow belt, red belt, blue belt.)

[PiggyWhiskey]

Expandable copper smelting (can be upgraded to electric furnaces later), optimized for beacons and (almost)blue belts. max is 54(missing 2) furnaces, but you can add extra with beacons.

~snip~

At the bottom is a 4 lane belt balancer.

That's quite a nice furnace layout.

I hadn't considered a single buffer per 2 furnaces.
My goto design was a combination of something I found on reddit for the fuel line, and Arumba's standard furnace layout.



It's quite expensive with the fuel underground belts, but when you transition to electric, these are removed to make the room needed.
It also allows the same width if you're after a compact smelting area. So you don't get the lane down the outside empty/having to move the setup across.

Also I added in the splitter with underground belts to allow a single line to output more than a basic belt can handle (Not sure if its efficient but I use 16 furnaces/8 rows before I shove that lane across). This (and another one later down the line) can allow triple the output, depending on how you handle the lane balancing/belt speed on the output. (It doesn't need to be red under ground belts, I just used them because I had some spare)

And I also use a faster belt on the merge point to get better compression. It manages to get enough of a gap to allow another set or two of items which compress when they hit the slower belts.

Also removes the need for long handed inserters....cheaper power etc etc.

I haven't thought about where I would put the power lines when I switch to Electric. I might need to double the amount of power poles using the basic ones. Or perhaps mix in substations. Not sure yet.

Whats the max furnace/belt setup for maximum efficiency?

[Tallinu]

That is exactly, what is described in the wiki: https://forums.factorio.com/wiki/inde ... s/Priority

If you use two express splitters directly in row it is guaranteed, that the both rows are joined together 50/50. Or you can try, if it works.

Thanks, the "Joining with Splitters" thread from that article had exactly what I needed to know. If single red or blue splitters do this reliably now regardless of placement, that'll be awesome, but at least it's possible to work around it with any of them. (Which explains why the design I posted worked - I must've done so purely by chance.) I hope they can get basic splitters to do the same eventually. It only should happen when both inputs are at full compression, really, and that should hopefully make it an achievable goal. (Prioritizing one input over another is what side-loading is for, IMHO! It's a much more transparent process there, and can be easily manipulated to do what you want.)

Small lane mixer with underground belts:

LaneMixer.png

Without underground belts, I suspect Tallinu's design already is minimal.

Beautiful! You have taught me something new -- I never thought of using an unpaired underground exit as a side-loading point!

My goto design was a combination of something I found on reddit for the fuel line, and Arumba's standard furnace layout.

~snip~

It's quite expensive with the fuel underground belts, but when you transition to electric, these are removed to make the room needed.
It also allows the same width if you're after a compact smelting area. So you don't get the lane down the outside empty/having to move the setup across.

~snip~

And I also use a faster belt on the merge point to get better compression. It manages to get enough of a gap to allow another set or two of items which compress when they hit the slower belts.

Also removes the need for long handed inserters....cheaper power etc etc.

I'm pretty sure that for a fixed item input rate less than the rate at which a basic inserter moves items (such as a furnace's processing rate), the electrical consumption ends up being identical for all three types. I read or heard somewhere that the power consumed by a fast inserter and the speed at which it works are increased by the same proportion relative to basic inserters. If so, the energy cost per item moved is effectively identical (they just move more items, and therefore consume more energy). If that's all correct, using a long handed inserter with a furnace shouldn't be any less efficient (pretty sure they only use more power because they work faster too, even though they're slower than fast inserters).

So you could avoid having to build and then tear out all those underground belts, and you could also avoid having to move the buffer chests and inserters during the upgrade, if you shifted the chests inward by one tile (removing the extra belts side-loading onto the main line) and used long inserters to transfer from the furnaces, across the aboveground fuel belt, into the chests. And from there, onto the single tiles of fast belt (which is a great idea).

I'm going to have to start using that kind of setup myself, now, instead of putting the coal next to the input ore - I've typically just been tearing down and rebuilding once I get electrics, often in a better place than my initial smelting that grows out of awkward bootstrap setups, but this would save a lot of work if I have a good spot to begin with.

[PiggyWhiskey]

~snip~
I'm pretty sure that for a fixed item input rate less than the rate at which a basic inserter moves items (such as a furnace's processing rate), the electrical consumption ends up being identical for all three types. I read or heard somewhere that the power consumed by a fast inserter and the speed at which it works are increased by the same proportion relative to basic inserters. If so, the energy cost per item moved is effectively identical (they just move more items, and therefore consume more energy). If that's all correct, using a long handed inserter with a furnace shouldn't be any less efficient (pretty sure they only use more power because they work faster too, even though they're slower than fast inserters).

So you could avoid having to build and then tear out all those underground belts, and you could also avoid having to move the buffer chests and inserters during the upgrade, if you shifted the chests inward by one tile (removing the extra belts side-loading onto the main line) and used long inserters to transfer from the furnaces, across the aboveground fuel belt, into the chests. And from there, onto the single tiles of fast belt (which is a great idea).

I'm going to have to start using that kind of setup myself, now, instead of putting the coal next to the input ore - I've typically just been tearing down and rebuilding once I get electrics, often in a better place than my initial smelting that grows out of awkward bootstrap setups, but this would save a lot of work if I have a good spot to begin with.

So I got around to a new world and I think I managed to understand what you were referring to with the Fuel Belt.
Like this?


Is inserting to the belt faster than to a fast/express side belt? I have noticed that my metal plate output line's are sometimes patchy.

[ssilk]

Keep in mind, that this might change in with 0.12! The belts physics will change. That was also the reason, why I hadn't updated the pages for a while.

[deemer]

My first post.

Since others reported having trouble doing neat oil processing, here is my setup, which I don't think is too terrible:

OIL

Oil Processing Setup

light-oil-cracking_cropped.png (896.75 KiB) Viewed 13320 times

It should be scalable. It doesn't make a balanced amount of lubricant, so you might want to add a tank to buffer the heavy oil and a pump to make sure it gets allocated to lubricant if you want the most of that. The light oil to petroleum gas arrangement can be shrunk one tile horizontally.

[Tallinu]

So I got around to a new world and I think I managed to understand what you were referring to with the Fuel Belt.
Like this?
~snip~
Is inserting to the belt faster than to a fast/express side belt? I have noticed that my metal plate output line's are sometimes patchy.

Yeah, that's what I was describing.
I haven't actually tested it, but I suspect that side belts do a better job of merging items than inserters dropping items directly onto a belt. I think if you alternated every other furnace's long and short inserters, you could have two buffer chests outputting onto a single side belt... and make the whole setup two tiles more compact in the process. I'll try to build an example.

More Betterer

There we go. Keep in mind that this shot was taken with a stalled production line, plus I set it up right next to a preexisting electric copper smelting area. With fast belts at the side-loading points (or express belts if you make it big enough that the front end needs the throughput of fast belts) I think it might load the belts a bit more uniformly, but that'll need testing.

[MadZuri]

rail

Double rail T-junction. You're welcome.

[Neotix]

That T-junction is one BIG deadlock.

[MadZuri]

That T-junction is one BIG deadlock.

Really? It has been successfully implemented on ColonelWill's MegaBase without problem. Maybe you have a bigger network and more trains than he does, but I kind of doubt it. He has something like 50 2-6-2 trains running around, but what would I know about designing rail networks? You are clearly the expert here. Here is the video where he implemented the design.

[PiggyWhiskey]

That T-junction is one BIG deadlock.

Really? It has been successfully implemented on ColonelWill's MegaBase without problem. Maybe you have a bigger network and more trains than he does, but I kind of doubt it. He has something like 50 2-6-2 trains running around, but what would I know about designing rail networks? You are clearly the expert here. Here is the video where he implemented the design.

I'd think Neotix has a point.

If 3 trains happen to arrive at the exact same moment heading clockwise (straight or to the right) they would each reserve the next block, while simultaneously preventing the next to take their position.
Especially with such long trains (2-6-2)



Edit: I mocked up a design. I think the sizing is a little off, but the issue is visible. I also manually placed the trains until the first stoppage.




The only guaranteed no-blockage rail setup is where all the signals are further apart than the maximum length train (1 Loco with 2 cargo happens to be about 15 blocks. Which also happens to perfectly line up with Big Power Poles)

But for all this, it'll only block if they arrive at close to the same time.

Also, I know alot of people hate round-a-bouts but I've found them to be pretty safe as long as the entire loop is in a single block (slow to travel through for bulk trains, but still safe)

I attempted using T-Junctions on a map, and gave up and deleted it because of signal issues. Especially in my depot.

[MadZuri]

No design is prefect (wait until we get those chain signals tho) and everything has trade-offs. Very large rail segments are incredibly slow, but like you said, relatively safe. I say relatively, because if the train count gets high enough, even that safe system will jam. The posted design is possible to make jam (easily solved problem with chain signals), but incredibly unlikely, and as a whole will allow a much higher throughput than a loop based system. Please do watch some of that linked video, Will's rail system is pretty amazing.

[n9103]

Personally, I make all my T-junctions without inner signals, and outgoing signals at least a train's distance from the nearest junction.
Tends to be slower on average than this system could be, but it's deadlock-proof and slightly faster than a full roundabout.

[Tallinu]

I just now needed to come up with an even 5-way splitter. Fortunately, I remembered seeing some even 3-way splitters a ways back (I think it was here), and I was able to figure out how to apply the same principle. Here are a couple of more compact variants of the result.



The one on the left would be good if your input is parallel to your outputs and you want to spread the outputs apart (more like this), since the right angle after the first splitter would make it easy to stretch that section out and divide the belts only where necessary. The second might be useful for evenly loading at a vertical train stop with five inserters, or other higher-throughput applications, since there's no corner on most of the lanes to slow things down. (I haven't tested these with a fully compressed belt. It could very well be that the first one handles high loads better due to the corner carrying a 50% load compared to the right one's first corner effectively carrying an 87.5% load...)


And in case anyone's confused how this kind of thing works: Basically, you figure out the smallest power of two that's equal or greater than the number of ways you have to divide your output. You split your input into that many outputs, and then recombine all the unwanted outputs and recycle them as input. When done properly this is guaranteed to result in the same number of items coming out of each output belt, due to the deterministic way Factorio runs.

[n9103]

When done properly this is guaranteed to result in the same number of items coming out of each output belt*, due to the deterministic way Factorio runs.

*as long as there are no item collisions on belts.

As pointed out earlier, it's theorized that corners won't matter with the next patch, but I haven't seen anything to suggest that this would eliminate the "accordion effect" that happens when one item bumps into another, which is what throws most balancing/splitting layouts off. The elimination of curves penalty will make the designs a bit more natural looking, but I don't see it changing the slightly different behavior of empty vs packed belts encountering side-loading and splitters.
But, eliminating those two penalties seems like it wouldn't be hard after the corner fix.



Personally, I'm not a fan of any of these belt "improvements" as they take away some of the quirks that make Factorio interesting, but *shrug*

[oLaudix]

I haven't seen anything to suggest that this would eliminate the "accordion effect" that happens when one item bumps into another, which is what throws most balancing/splitting layouts off.

Afaik items on new belts are not individual entities anymore, so they dont have collision box. Not 100% sure i understood it right though.

[n9103]

They talked about lightening the processing load with something like that, but I don't know how it'll actually work, or if they've even solidly committed to that approach.
Something I'll be keeping an eye out for in the next release. Can't say much about it until then.

[ratchetfreak]

They talked about lightening the processing load with something like that, but I don't know how it'll actually work, or if they've even solidly committed to that approach.
Something I'll be keeping an eye out for in the next release. Can't say much about it until then.

they snap the items to rails now and that means they have to check a lot less collisions for movement, though I don't believe it will change the accordion phenomenon

though IIRC most compression loss is from corners and the rails physics will solve that

I kinda hope though that the rail on the belt entity will be moddable so the rails can take many forms to let people create lane swappers and such

[Tallinu]

When done properly this is guaranteed to result in the same number of items coming out of each output belt*

*as long as there are no item collisions on belts.

Right, that's what I meant when I said I wasn't sure how well it would work on fully packed belts.

As pointed out earlier, it's theorized that corners won't matter with the next patch, but I haven't seen anything to suggest that this would eliminate the "accordion effect" that happens when one item bumps into another, which is what throws most balancing/splitting layouts off. The elimination of curves penalty will make the designs a bit more natural looking, but I don't see it changing the slightly different behavior of empty vs packed belts encountering side-loading and splitters.
But, eliminating those two penalties seems like it wouldn't be hard after the corner fix.

Somebody got hold of a video from a dev showing an example:

https://www.reddit.com/r/factorio/comme ... new_belts/

Looks pretty damn smooth to me. No accordion effect that I can tell. The reason the left and right lanes become misaligned is that the inner lane on a curve has less distance to travel but goes at the same speed as the outer lane (even though that's unrealistic behavior). The path includes four more left turns than right turns, so the left lane ends up ahead.

This is going to not just eliminate slowdown on inner corners, it's almost the opposite, which mildly amuses me (although the throughput is the same on both sides).