Steel Furnace setup: 3-express belts of output

[dragontamer5788]

I do realize that most players may prefer electric furnaces at this stage of the game, but electric furnace builds have already been studied in a significant fashion. I felt like the status-quo for steel-furnace setups was lacking, so here's my contribution.

https://s3.amazonaws.com/dragontamersvi ... erview.mp4

It was surprisingly annoying trying to get all of the compressed outputs just right, especially at the end of the line. In any case, the 3-express-belt magic number is exactly what is needed to load and unload a train (12-stack inserters output approximately 3-express belts). I ended up having to use belt-weaving for the middle lane due to lack of space.

In any case, coal is very plentiful and steel furnaces are a lot cheaper than productivity / speed modules. It would only take roughly 10-electric drills of coal to run a setup like this, or only four solid fuel plants. As such, a build like this would be useful in perhaps building up or bootstrapping a megabase (at least while you're still saving up productivity and/or speed modules)

-------

I personally find it marvelous how 4-express belts of input turns perfectly into 3-express belts of output (two of the lanes are taken up by coal. The six other lanes perfectly compress by the end of the setup). But... those are the silly things that interest me I guess.

[siggboy]

Whoa, nelly .

You like to make things complicated!

I also like steel furnaces until very late game when there's enough energy to go around to make compressed setups with e-furnaces and beacons.

However, steel furnace columns are one of the easiest things to build in Factorio. I'm wondering why still so many sub-optimal designs are posted.

One blue belt = 70 furnaces = two columns of 35 each. Split the input, output into the middle, and for the last few rows you need to output directly into underground belts so the inserters won't clog up.

Done. Make as many of these as you want for 1 belt, 2 belts, 3, ... 10. A red belt of coal is enough for almost 17 such furnace columns (17 blue belts of output).

(BTW it's not so good to upload a link to MP4 files; most people won't bother downloading. Better post a screenshot and use LICEcap to quickly make an animated GIF: http://www.cockos.com/licecap/ )

steel-furnace.gif (7.24 MiB) Viewed 36879 times

blueprint

Code: Select all

H4sIAAAAAAAA/62d3W7bOBCFX8XItV3oz46Dwje96AvsAxRuohYGHNuwFaBFkHdfJVJ2I4szOnNOrhZIlqQ084nl4RkyD8fZ/ni/3c9+bJ7rQ7NrdvVl8/x8
2D7Wm5v6z+lcXy6L5rw9XE7Hc7P4We+bm/npeGn/x+Nh8/xnk83/bhbl3cv8YXeu799+Wr3M+w5+bS+N23qxfGu+/ti8CDQvteaV1ryYbP50eKjPv8/H9r+p
DvJEB/Pm76ltvjucnpqbSG9OZ8enZtAbnth1OrF4B7ckGVpmy8mxL6f9rmnq87Bdl9HbL0tyXA1nF0ck5P3ntOST1n8Sgx5WsR5KuYdC7iFXe0h0EEtFhjxB
GsLsS5fGlYOvPxWUifbpeQV5lURCY7FI5DPWgYx1gmrqyyqFL6vroUr38HP3e1Hv2x+fd/eL03FfX38SPRNlOzUhE9k7Q9ULPv3k/SviLYpxUFbRSdprjVBe
pShn/sGTM+wmOIBIIfeQp3vYHS71eYRKP+Xn6TxcmrreL349nQ/bewvK/COUxiC5N0i6TaIJMwcbwXBeLPBeidgxU5vxjHgHMjOFsYQPUJcR1GU0dcUdTJ0x
iEddJlKnhDPzojmN7TAwMELWE3vcC1FykQlgKwiUvgdDJXjYDtvEsL39mJ3H+mH39Oj+y5/3rwmhbj2Yg/qwCYO6qhGtDACoB4NZXccS/zqMZ8Q7MKLkfF5K
Znq2FSXY9WCIEPfrWPFfxxKd1K1BPNJXKulCODMvmgDpS25SF56YyH8i/RS28hq6MJbxLraGEkGwLWFsjUE8bEsVW0HTZF4+AGxLDlsBACL/ifRT2Aqr+L4H
YxnvYlvw2MLCzRrEw7ZQsRXCmXnRBLDNOWxDujuRPWIJbb1kADp5EZ6HpEOH7bBNCNv8jlpCZxjq1oM5qA+bEKhbAcRRp9ViNJjVdSxxUkP7AwlIiK9DCWzP
trDw73sIrfz7r2PNfx236KRuDeKRLphTmRrOzIsmQPotNalbA+IdhDYYEuln1iK5sW4P9GBIBxdbw+9CsIWVnzWIh61q5FnBwDuglR8SmETwmclSFYufAJ1s
sOWGdHCxrXhsYeVnDeJhW6nYqsrPiiaALaf8rAHxDuLKTwlzD528CM/jLtV/CCIqrhoP8j+C6BK6fU0M9bjNl6s2n5KCjMxAMgHgEno6lomUMZO6ai1+Atuy
wGT0Je8swsaiNYbHuWosKsHMnFhOY87ZitbzesyrtqIs+mRTMe5ovcOHqLdqNMQUex2vcXNQ9QZVa5B2BjmtF/fohAB5oOCkyQ4fYfDx/h4s8uLunmruqQqP
FnicsydILY8bB3XZ1tMr4+Ko8p5eyWwyVxjdcRNQ9QBVIUjrwGAYq6sowjyHZP8YDaZ4SV4Xy84N4RbyOx2wVxi3CtU1seqe0MoP3JqTTT5Vw4mkqQIuXo9M
VyNnIKbhLQpxg0JklPX3gHCMwh0HVODDwQPGS52J4/MoPY0WIJ7hSVScQ8X1ATuDAuEYhTuOp3imQ90AUyVa3NGgbbiKWPRia96waSdadqKcY9e7wQBWw/ih
RIomSVj7qQfn1F2O+B4FvUWxAifo8AaFuD8h1k2wuxNAOEbhjk/Q4maGipe63RvfUu9hW4ObvaGStDc8w0VsYgmbuNHLmhJrZu9MLOgJn5MSQtupJ3V9HPNg
Ov1F22c5rMDC7plqnimBzJw4TmI6jAksi8LlxKpzJju1cv2O5TZ7qNJ7vHlBLHfBIgjrqRy6xT1etQLCyt003cEwVldRhOkMl/YoIfVoxHFWPQ+rJg+AuwIX
F7HCtA7ucC2bWsqmFmDShWzDMMLgiAotVnw7ziKzylCLN2N1zh169DnqHNVp1hAOqmLRg1riTpcKD2MCoyaebLYe10FVPdcsn+ogjmTwJzJyROXlzhAOquJx
DPXcEX0YI6eEm/K48bSPs06gKh+ui50p7MwJ2jwrMmJBDB6ss57qPRL+XU/ZOBTpC80ch0R08OgzdtGoVldBhX2KcKmlEhIPTpxu2bojvDvevIPdu7D4mzp9
DnwdZerruL4IbRrWgpmZlTzG0zjOIjMzq7acdUmBhx5tzBWImstHTwWhN6EcAPQqYGKeJo9SWlYaHXLEc3Ly7STKXUJdB3GF8I4RIpuq0RDho0JtcwjWsDib
uhMJgPX2c+ZJSmzFLhsbJ4KgVZX18g1rRfz0TmHYVwDgqNgqwmKrmNACAHrrT5kn18RykrjzLJa3cdoIVlXUlHs0uw7iThl/hWWJOmXxa1OnbiSdZrVMiq3w
NFlSzlfsAttxFogSMVWny1dDW1e4eujRzlfJOF8fr4b2aA3XLpYTa3mA1uIzZtYS0HfjuBOwifXcOmuq9rFuuQbIQ22pMiRkOvLCRYZTV8ID5C0/Z54kCgdL
DFbBGFJZ8VDBWYtX/5WGegHwRH0lawiHtYmVPMAa8hcdplGjlEssC+MkUO97531bL/Pd/fHw+teKLrvfh+1+89zHo6kfb+apYLQt2jHbbLRtu9//8/r72ffu
97N89m3/VM++vT7My9dz3bQ/n/34Wh8e/gVE0EekJ2kAAA==

[dragontamer5788]

Whoa, nelly .

You like to make things complicated!

I appreciate the criticism!

A few things before I go into your post: 3-blue belts feed 12-stack inserters very well, which means you can load a single wagon. So anything less than 3-blue belts will inevitably have to scale up to this size (at least!). "Keeping up" with trains takes a huge amount of effort, as those trains move a lot of plates very quickly ya know.

I was very deliberate in choosing 3-blue belts (120 plates / second) as my "target". I will create this steel-furnace setup for every wagon of ore at my train stations, which will turn into a wagon of output.

However, steel furnace columns are one of the easiest things to build in Factorio. I'm wondering why still so many sub-optimal designs are posted.

One blue belt = 70 furnaces = two columns of 35 each. Split the input, output into the middle, and for the last few rows you need to output directly into underground belts so the inserters won't clog up.

Done. Make as many of these as you want for 1 belt, 2 belts, 3, ... 10. A red belt of coal is enough for almost 17 such furnace columns (17 blue belts of output).

While your design is simpler, I disagree that it is optimal. Consider that your design only outputs 40 plates per second, while mine is 120 plates per second. If I were to scale your design 3x larger (and then steal some compression-tricks you're doing... nifty underground belt setup btw) I would bet that mine would cover a smaller footprint.

If we scaled your design to 120 plates/second, you'd have to run 9-express lanes down the full length. 6-lanes of input, 3-lanes of output. Half your belts are coal, and only half of them are the input / Iron Ore to be smelted. In contrast, I only have to run 7-express lanes across the track: 4-lanes of input, 3-lanes of output. So I'm saving roughly 420 blue-tracks (and therefore 420 tiles of footprint) compared to your design. (35 * 3 (triple the size) * 2 unnecessary express belts * 2-tiles per Steel Furnace == 420 unnecessary tiles at 3x scale)

On the other hand, your design uses the underground-belt trick for compression, which is something I probably should do. If I steal your trick, I can remove the runs of belts and splitters I use to compress my output. Thanks for posting your design, lemme try to steal the best parts of it.

(BTW it's not so good to upload a link to MP4 files; most people won't bother downloading. Better post a screenshot and use LICEcap to quickly make an animated GIF: http://www.cockos.com/licecap/ )

Thanks for the tip. I'm new at sharing. When I get my design updated, I'll do this.

[siggboy]

While your design is simpler, I disagree that it is optimal. Consider that your design only outputs 40 plates per second, while mine is 120 plates per second. If I were to scale your design 3x larger (and then steal some compression-tricks you're doing... nifty underground belt setup btw) I would bet that mine would cover a smaller footprint.

I would bet that you would lose that bet .

The design that I've shown is meant to be tiled, you can place the blueprints back to back. There's simply no room to save any more space.

(Actually I've assumed that it's clear that I was only showing the blueprint for one column, it's easy enough to extend it to 3 columns simply by repeating it...)

The blueprint that I've posted is 11x77 tiles, so for 3 belts you need 33x77 tiles = 2541 square tiles

Your design is over 130 tiles long and 23 wide (I eyeballed that from the video) = ~3000 square tiles

So your setup is about 17% bigger, AND it's not tileable, AND it's more expensive to build (more splitters, more inserters, fast inserters where you'd only need normal ones, etc. etc.).

[dragontamer5788]

So your setup is about 17% bigger, AND it's not tileable, AND it's more expensive to build (more splitters, more inserters, fast inserters where you'd only need normal ones, etc. etc.).

Ah, but the fundamentals of your setup are inefficient. You saved room because of the underground belt trick, which I didn't know about before this topic was made. Again, when you scale to 3x, you will have 6-lanes of input and 3-lanes of output across at least 70-tiles (the 70-tiles next to those steel furnaces)

My fundamentals are 4-lanes of input and 3-lanes of output. So my fundamentals are at a 420-tile advantage over yours. The reason why your design is more efficient is because you're a more experienced player who knows more tricks (in particular, you use very little space for compression. I on the other hand, use several splitters to compress the output). If I steal your tricks... in particular using that underground belt for compression... I think an updated design of mine will beat your footprint. Its something I'll have to lab out for a few hours again however.

So your setup is about 17% bigger, AND it's not tileable, AND it's more expensive to build (more splitters, more inserters, fast inserters where you'd only need normal ones, etc. etc.).

Its tileable for sure! There's nothing outside of the "outside" lanes.

The blueprint that I've posted is 11x77 tiles, so for 3 belts you need 33x77 tiles = 2541 square tiles

I'll be back when I beat this. I bet you that with the 4/3 setup, I can probably get down to 2300 or lower.

fast inserters where you'd only need normal ones

That's convenience. I've got fast inserters all over my factory because its more convenient to carry around 800 fast-inserters rather than 400-fast inserters + 400 yellow inserters.

[siggboy]

Ah, but the fundamentals of your setup are inefficient.

The "fundamentals" are inefficient? How? It does take 3 belts of ore and turns it into 3 belts of plates.

You saved room because of the underground belt trick, which I didn't know about before this topic was made.

Yes, and you not knowing about how to design efficiently does make your design efficient? That's some twisted logic at work here .

Again, when you scale to 3x, you will have 6-lanes of input and 3-lanes of output across at least 70-tiles (the 70-tiles next to those steel furnaces)

No I have 4 belts of input, 3x blue belt of ore and 3/17th of a red belt for coal.

(By the way, my design can be made 2 tiles shorter, for a footprint of 11x75 for one column.)

steel-furnace.png (793.88 KiB) Viewed 36841 times

My fundamentals are 4-lanes of input and 3-lanes of output. So my fundamentals are at a 420-tile advantage over yours.

I also have 4 belts of input and 3 belts of output (see above).

The total footprint of my setup for 3 belts of output is 2475 square tiles.

Its tileable for sure! There's nothing outside of the "outside" lanes.

Yes, you can tile the entire thing for 4 belts, but you cannot expand it belt-by-belt. That's what's usually meant by "tileable". (Otherwise the entire factory is "tileable" in a sense.)

I'll be back when I beat this. I bet you that with the 4/3 setup, I can probably get down to 2300 or lower.

The number to beat is 2475 square tiles, not counting splitters before the input and after the output (with that I mean splitters OUTSIDE of the screenshot posted above). Good luck.

Don't waste too much time .

fast inserters where you'd only need normal ones

That's convenience. I've got fast inserters all over my factory because its more convenient to carry around 800 fast-inserters rather than 400-fast inserters + 400 yellow inserters.

Yeah, absolutely. I also don't use yellow belts even if it's possible, it's just too much hassle to carry 3 colors of everything. And end-game I usually only use blue belt even though that's a colossal waste in many cases.

[gheift]

Just want to share my design: same footprint as siggboys (if the final splitters are ignored), but much cheaper:
roughly 240 belts + 70 underground belts ≅ 960 iron-plates vs 220 express belts ≅ 4750 iron-plates.

Furnaces

Three yellow belts as input:

furnace-yellow-input.jpg (38.25 KiB) Viewed 36834 times

One blue belt as input:

furnace-blue-input.jpg (332.75 KiB) Viewed 36834 times

[dragontamer5788]

(By the way, my design can be made 2 tiles shorter, for a footprint of 11x75 for one column.)

You realize that my design is only 23-tiles wide for 3-output belts, right?

Your design is literally 50% wider than mine (33 wide vs 23-wide). I'll need to be a bit longer, but I really do think that reducing a few belts will lead to savings. We'll see once I integrate a lot of the efficiency tricks that are in your (admittingly well optimized) design.
-----------

You've got a lot of efficiency tricks I didn't know about. I'm integrating them. I ran out of time for now, but I'll be back. I'm very confident that my solution is a leap forward from yours. I think you're not seeing the inefficiencies in my own design that are easily fixed (now that I see your design anyway)

[siggboy]

Just want to share my design: same footprint as siggboys (if the final splitters are ignored), but much cheaper:
roughly 240 belts + 70 underground belts ≅ 960 iron-plates vs 220 express belts ≅ 4750 iron-plates.

Yes, that's the same design principle, only shorter columns and then merging. In early game I usually built a short version of this a few times, also for the speed-run (8 hour achievement) this is a good design I guess.

It's nice because it allows you to scale up. The cheaper manufacturing cost is great, too, but not such a big deal at end-game when you can make everything in bulk (as said above I usually end up using blue belts for everything just because it's easier than having to carry around three different belt colors).

Probably steel furnace designs is not something you should overthink or make a science. But we'll all be surprised when dragontamer dazzles us with his 2000 square tile design for 3 belts

[gheift]

Probably steel furnace designs is not something you should overthink or make a science.

Why not? That's why I play this game: optimize the hell out of every design

By the way, I have a few other designs:

• 1 blue belt: 657 tiles
• 3 blue belts: 1971 tiles
• 3 blue belts (max width 23 tiles): 2240 tiles

Now you have something to beat

proof

Code: Select all

sha256sum furnace-compact-2.jpg 
ea89fd7f000cd113cd1d634785770e6c5f6d6026f33b03d01afd247d569eaa3c  furnace-compact-2.jpg

[dragontamer5788]

I admit defeat, although I have learned a bit about optimization. I have virtually tied your design siggboy, but alas, the slight inefficiency eeks out a victory for your design. I've dropped my design down to 23 x 107 footprint == 2461 (for the furnace section, not counting the tiling setup ahead). Your tile setup is 33x70 == 2310 for the same comparable area. (Furnace only, ignore the tiling portion).

optimized.png (1.24 MiB) Viewed 36513 times

I still think my design has optimization potential (I might be able to tie your footprint), but I've run out of the "big" mistakes to fix. So alas, I have to admit that I'm wrong.

I regret nothing. It was a learning experience mimicking your design, especially the belt-compression trick (which I'll use from now on). Nonetheless, I still see inefficiency in your design.

count.png (709.08 KiB) Viewed 36513 times

Despite you claiming otherwise, you clearly have 6-lanes of blue-belts running through your system, exactly half of which is coal sitting around not doing anything useful. Your coal lanes are designed for 120 coal per second, and yet this system barely uses 5-coal per second. Clearly, reducing the bandwidth of coal while increasing the bandwidth of iron ore (proportionally) should have yielded some savings.

Yes, I realize you are only feeding your coal belts a red-belt. That doesn't change the fact that you have 6-lanes of coal on blue belts running north/south. The entire point of me approaching the 4-belt input methodology was to "reduce coal bandwidth", so that I could gain those last few blue belts and improve the footprint of these systems.

---------

I think my mistakes were multiple fold:

1. My math was clearly wrong earlier. Very far off, yeah I probably deserve to be made fun of with some of those mistakes. Whatever, I"ll learn.

2. Now recalculating the math, reducing from 6-lanes (aka: 3 belts) of coal to 4-lanes (aka 2-blue belts) of coal should yield savings of only 70 at max. And most of that has been eaten up by minor "mistakes" in my design.

3. I neglected the 50% width penalty of my design (57.5 furnaces wide instead of 35). The .5 is especially worrisome because the lack of a whole-number will only create more inefficiency.

mistakes.png (1.1 MiB) Viewed 36513 times

3. I lost 23 tiles by leaving a hole to merge in the ore.

4. I lost at least another 23 tiles on the "17.5th" steel furnace. You can't have .5 of a furnace on the line, so I'm losing out on the width at that point being 18-furnaces wide instead of 17. I recycled the empty-space to balance the belts of output, but the 11x35 design doesn't need rebalancing.

5. I clearly have "lost" more tiles than that, but where and why?

-----------------

As such, I don't believe that the "perfect" steel furnace setup has actually been discovered yet. An ideal design would be better-able to take advantage of the excessive-coal bandwidth and cut it down somehow.

Probably steel furnace designs is not something you should overthink or make a science.

Why not? That's why I play this game: optimize the hell out of every design

Also, basically this. If you aren't optimizing your designs in this game, what are you doing exactly? I mean, there are literally infinite resources. You could have a horribly inefficient factory and still launch the rocket. In any case, I don't regret coming up with this design, even if its less efficient than siggboy's. I only wish that I were able to figure out where my design goes wrong. (IE: Based on my design criterion, it should be only 2240: 70 tiles less than your 33x70 design. Why am I instead 150 tiles larger?)

Perhaps if I instead changed Siggboy's design so that the middle lanes: 2 and 4 to be 100% Iron (instead of coal), that would get a 6-belt to 4-lane output. (If perfected anyway). I mean, all I really have to do is add red-inserters and then do the rebalancing that I did in my design. Then figure out what to do with all the extra iron at the end, lol.

[LoggerM]

Hm, I asking myself: what are you doing?
You need a big Fabric to generate enough Materials to generate this lane.
Presumably you need much more Ress for this setup than starting 1 Rocket.

I don't need optimations for Space, because we are Playing at an endless Map, right?
For me its more importend to optimize things for Setup-time (for speedruns) or required ressources at low-Ress Maps. May somebody can do this for us?

[BlakeMW]

coal lanes are designed for 120 coal per second, and yet this system barely uses 5-coal per second. Clearly, reducing the bandwidth of coal while increasing the bandwidth of iron ore (proportionally) should have yielded some savings.

Using rainbow belt technology it is possible to solve this problem. The basic idea is you generate a count-perfect mixed belt of coal and ore, if my calculations are correct the perfect ratio is 200:9. How these mixer belts work, is a counter keeps track of how many ore vs how many coal has been allowed to enter the belt, and it maintains a 200:9 ratio by switching the belts on and off. When dealing with a very skewed ratio like 10:1 belt mixers are unbeatable for compactness.

There are a few problems: The first is you need overflow, the setup must be free-flowing. This may be a return belt, since underground belts can go under inserters it doesn't make the setup any wider (also it can be a basic belt, no need for speed at all).

Secondly a naive design can only produce a 95.5% compressed belt (ideally) because 4.5% of the input belt is coal.

There is however a solution which allows a 100% compressed belt, and it isn't any wider as it exploits the ability to place mixed belts under inserters and merge them later (technically you can always merge belts, but you can't use this trick with 1 lane of ore and 1 lane of coal, it only works with the count perfect mixing that lets a single belt perform nearly double the work of a normal belt with 1 lane for each material).


A single strand: In normal operation designed to be tiled and share output belts with the next strand.

The trick is this, we initially mix the ore and coal onto two express belts which can easily accommodate the ~1.05 belts of contents which are needed to produce a compressed output belt of plates, those two belts are merged later down the line. This gives the furnaces access to an entire belt of ore.

The mixing logic needs a slight modification. A splitter merges a compressed belt of ore and a compressed belt of coal, normally with a count-perfect mixer you only let one or the other enter: In this case though, the splitter outputs to two belts so there is room to accommodate both input belts simultaneously. As such iron is allowed to enter the splitter "unconditionally" (under normal circumstances) and the coal belt is switched on as required to maintain the perfect 200:9 ratio of ore:coal. Important thing: There is always a full belt of ore entering the splitter and going down the line. Sometimes coal is added but this is not at the expense of ore throughput.

This system generally speaking consumes all the ore and coal which enters it. However it must be free flowing to accommodate "runs" of ore or coal caused by inserter bias. In this case, I bring the overflow back to near the start and just sideload it back onto the belt, exploiting the gaps created by the first 4 inserters. Under normal operation overflow is practically nil so this solution works fine.

Zoom in:



Wiring:

Combinators:

• Pistol = 0: Output A (resettable counter)
• Ore * -9: Output A
• Coal * 200: Output A

Belts:

• Ore belt: Enabled: A > 83500, Read Contents: Pulse
• Coal Belt: Enabled: A < 84000, Read Contents: Pulse

Why 84000? The system needs to be primed with coal because each furnace will buffer up to 6 coal but only 2 ore. So this means you need to start out with a large influx of coal to fill all the fuel slots. The amount of buffer coal (420) then needs to be multiplied by 200 (the skew ratio). 84000 is slightly rich, it results in a small amount of excess coal circulating along the overflow, but it's better to run slightly rich than slightly lean because an unfueled furnace stops consuming ore.
83500 is the trick that allows iron to always enter - basically unless the coal is more than 2 behind (and that can only happen if the coal belt dries up) the ore is always allowed to enter. If the coal belt does dry up, the ore belt is throttled as appropriate (and vice-verca).

Another thing: This system breaks down somewhat if the recirculating belt is allowed to fill up, this only happens if the output is blocked for long enough for the furnaces to fill up. The easy solution to this problem is to detect when too many items are on the overflow belt and just cut off the the input.

Here's a blueprint, it's not identical to the screenshot as it includes the wiring required for dealing with blocked output.

blueprint

H4sIAAAAAAAA/9VdbY/bNgz+K0U+DklhyrFzh84D9juKIfAlup6BxA5spejhkP8+x85d61im+OJi2L4MvZiiSD56SFGKs68+Hapdfvi0zd5s6QpX2CZ7eyvz
o80Wu6psXF661a46PhVl7qp6sTxVTftUVWZvP7JV/LB8zVbJZbkvarvr/pwuWzFXV4ftk33JvxdVnb09Fwdn6+vATfGtzA/Zm3s9tQoKZ4+LZa/sVDSuOiwu
rfi5dBksi3Jvf2RwuSyLZtsO/JwfGnv9uCx7Ve14Xxew+Cd7q+2+Hbub/+u22Gfp43JX1Ltz4a7/Mpfrf8ubUfbHqbZNs3J1XjanqnarJ3tw93atR3YZ5gix
egSjHgHUI0TaEcyjeoQxxrgjbNQjpOoREvUIakwaNaKMGtVGjUmjxiSoMQlqTIIak6DGJKgRBWpUgxpRoEY1qDEJakyqIalGpBpOakSr4ajnWHXCVCNJOYAW
iFoPaOevxYA2ANpVoIWgdhlrF6GaybREpqZzNZur06I6p6lTuzovqwscdX2jLvNIVd653WTW3+p2y7n3zcGDxeVt61qezm7Bs4gCzMCEPMDAJxQYz4OT23jV
2QkG9G0YVRP0BDEwQTQ7UJZ2wELP4ta5bDqmwgGngypArafFw8zIGBPs7a5obQl1mMzn5OLpK92kt+0n+06kufaa6sZtSX2mvrvVlkyt9lNeX7Vni2yx7N1+
N8b3onbn/PA+TP/h6u9uoNPrtutabZ/r6rjtnJy5+mwvsnbV8pfPIFlfm1dfFyYsCUPJwb820XRLLK8L93K0rtghcdggcfg5ACsURV2Vq6q2vwaj3QhVJ9vG
4qp48Qc3GESPc7w6jMcGlH5M5/fjrsoHgDZR9N87EQNc46w9rJ7PdZnv7Khx+jl5d9GHQFE2tnZ25MyOzte/UsuapgQ8SjAB88iYFSCzmpKJhJaYjWdiR7sv
zseVPbRD1S0aT9VhbFCvkKYkoVvfN1t51vdljcR6H1hQAV/gp2YVCywxQkvABzBKHCNGHMEHlglL4JFvfb9rkFjvAxgqwGAJwFhiSiaRWuIDGCGOYBhxZJAR
YMQyJYMRGDYvH7yw5xm0ImAVKan4oEWI4JqzEBksJEgmMsMjXvQM2QZ+PhSy6FoUOUbgUrLNGPj8EkLGeeBFDXxR9k+InwBQssEm5cNSOHAczgQfNias4Od9
NMVgs/IBCnveF+2JKWG5eEIEI1mMCXyYCoePU7oYOtsYAd0I+cb4QIU9T+cPtCSeEBEyiPGBihC+B3r4YjrrGD7tSDdPsQ9U2POy/BLfJ5gJK4RpO/aB6jlv
3GpCD4YsTA6DF+1wtEMMaI5XN+MRPH2U98bDRxPF30PBWnq3VspDnETD/uBf149v49syfzrY7b5orv/vWn4fn9U2329f8nJ/nYWzpWuGn7//tX/wWO1tFkna
hXdtPkA+i6Y/M1HivSXXnA6F81SYHYaiz6qj8uT3x9J+t/WreynKb78GNR4E9E95QLsriTNE1ETpwPuE3UbvPTOOGH42cGuZ3XndeziAUMHtdiOwOeR2qVEi
GCGCRNuNh380h1O3Jpp3wPujG8w2M2Y1mlPMeAkRBcco4HtzjAHdudWteabFZl//CZwC42gSBce4YnsTfJlN5c3+EoHWmyCFGEgpAjCKCOSlfgMazEshz40R
JFnWGCowOSkMx/GmyWFcgMhhhIzGVyYnTBxCpwhRLwy5B3A0OaFTQBhyD0nR5IR+MUKMeVITTU7oFyNEmad+oG3FhH6RVl+x0D5PiS/ZMkZ+NqdUyVdx4H75
ZzPS+Xv3mOvobo+p2JLMtMfsz/a5jkt+u+P+Jxs6SNLh9hqG22t0p7OWYHY9vU7o31dTDWCmB0DNBYm5IFQWSZRFKg7qSB7ojYABbgzwl6FBOBPdOEnIsrNO
gxszZlvmAOkcMwDffawJFhmyhICEqWi4Y43Q9jnt7fjpiFRxVbZvLqj8inBSsDkBNLLsJkl91gyfpazdWLIkECakf8tyYoAQBiIqBkIDwdRAgjvE/bRULhES
PjzQ8QGPgmCDlv8AYWv690Alrkkl5iJsTbnltBnqDDVi1rMyGiCMFppJMudyAG21BsJiC4wk5ggdBTuA1LUXsbkZJIUjIDwSgMAjEQGBYR5mhbSWfYSlYkqP
7EYQJW1Fpy0phYWLZOOGUBJhDSQsOjXz1ofaDaN4JcazcrEo1oJ9JFIChdrlLGIU0KIylDIXGg52yYVA4KtxMyJH2TBRlgBKikRoJ3AqQqN9QW2JkHbgxIWp
SNVSFqQzJEfjaIWZNnPUyoX8fgB+mCSFJ1arhQ7TuKq0DIiQAYGxBV0GbAsROvkjsy7M2lHCNqrklxrIFtKcuzdsCxw6O+UGWVaeM5oeRlDIaFseWDOIsFgE
XRqsIxd4/8CcbTCsiUV4EcVsCxHrL4ZOx4mwEjAa1i4myau2UveNZJJGWefNpBwOfj+dmAeBCHcFpjFr3wI7oiLJIxQUujfBDDJ2+he62sFVhfADKifrBsbs
85D7s1p0UuL9fUze7AVf9SLTT90lhm6+0LgyThlORRgHlVOe5XmuxND0jlmDLjfQt6YGj0qXhHdQkL/HfHudBPdtEvTnN8znE+bzTHsN015g2gtMe4FpLzDt
Baa9THOZ1jKN5drKejxiPW1YT69ZT6espx9YTwPPTODZCTxDgWcp8Ew1zIjyTDU8Uw3PVMMzNeaZGvNMjXmmxh5T6Tf/Jr7vgN0wXHP1xNN6KDdxY4lOM62T
fk+Q6RyQTDTi64kEerpdhfpl+up36fMs7UTYlqZ8PalET6J1CLL+6Le9mJZKFpNBFhN2Y46tB1lz9Mti6vf78yztRLiWArKYsPtobD3ImqNfEVO/559paSKx
lJ/MQJLMAFlz9AtY6t8cYFoqyUzAz0wgyUzaxKTNS/y0JMlK/KQkyUnalKTNSPyEJMpH/HQkwb/OF8pUxF7kgiXO9qPAjUq6VAJSmb3YuetuzU7fhWF6Uclx
yiqAnwkkiYBdegkKL23yxXIv+e4Fz5OCIgWrUZAX13G1KBcXVhmSrzjwrBSU0FgFjVzX4DKxNlsQiijCPQHKm/+R9+BxjWbnODOR5Ag/faEwbaIQI/w+Bumn
GZBX8rG1erpJHFOnOkuEX9rQmCogcqx3SXTVmCRZs55qSxJ+soOkNXTuy1c9iUjm4TQ3VMriJUYYmHDx4MbGl2Wx674xi/yQ7LDJfvn4HdklIjNhwbvw9dcL
vtTWtWN+2n6x5f5fGnJpYjd3AAA=

Btw I don't claim that this is the best layout for steel furnaces, although it definitely makes very good use of express belts. It is not the most compact layout, you can be more compact by using underground belts in extreme quantity.

[siggboy]

By the way, I have a few other designs:

• 1 blue belt: 657 tiles
• 3 blue belts: 1971 tiles
• 3 blue belts (max width 23 tiles): 2240 tiles

Now you have something to beat :D

Well, I've said I'm not that into making furnace designs a science, so maybe I'm the wrong person to pose this challenge to :).

However, I'd be very interested in seeing your solutions, because these seem to be very clever layouts.

(Hashing the image for proof was a funny move, BTW :).

Using rainbow belt technology it is possible to solve this problem.

I love that setup; it's hopelessly overengineered, but beautiful. Probably the best way to beat gheift's "challenge" :).

[dragontamer5788]

There is however a solution which allows a 100% compressed belt, and it isn't any wider as it exploits the ability to place mixed belts under inserters and merge them later (technically you can always merge belts, but you can't use this trick with 1 lane of ore and 1 lane of coal, it only works with the count perfect mixing that lets a single belt perform nearly double the work of a normal belt with 1 lane for each material).

Nice solution.

I would expect that the "extra 0.05" belt can be a yellow-belt instead of a blue belt btw, to save even more on blue-belts. That splitter at the beginning would basically fill-up the yellow belt (~13 items / sec), while the central blue belt would take the rest of the 29 items per second.

Because the footprint is the same size as the simple, tilable 11x70 design, I would guess the "real purpose" of the rainbow belt solution is to save on blue belts rather than on space.

[BlakeMW]

Because the footprint is the same size as the simple, tilable 11x70 design, I would guess the "real purpose" of the rainbow belt solution is to save on blue belts rather than on space.

It's actually a tile smaller when tiled because it shares a belt with the next strand.

[dragontamer5788]

Hm, I asking myself: what are you doing?
You need a big Fabric to generate enough Materials to generate this lane.

Not really. 3-compressed belts of input is just a single train wagon.

Imagine the 3-lanes coming from 24-stack inserters pulling off a train (12-pulling from the train directly to a buffer, 12 to more to pull from the buffer onto a belt), and you'll immediately have 3-lanes of compressed input. Similarly, the output can be eaten up by an inverted setup: 12-stack inserters pushing into a buffer, then 12-stack inserters loading a train.

Presumably you need much more Ress for this setup than starting 1 Rocket.

Nope. I'm counting ~650 blue belts in my (still imperfect) design, 210 steel furnaces, and 420 inserters. Pretty cheap all else considered. A rocket on the other hand requires 10,000 steel (aka 50,000 iron) just for the low-density structures alone. I'm estimating another 39,000 Iron for the 1000-rocket control units needed. That's why I'm interested in these Steel Furnace setups: they're so cheap to manufacture! You can do something like this easily pre-rocket, let alone post-rocket.

-------------------

But yes, I'm definitely post-rocket.

The ideal purpose of a Steel Furnace setup is to bootstrap yourself into mega-base size. You have the resources to launch a rocket, but you don't have hundreds of speed / productivity modules yet for any of the truly endgame setups. The few productivity 3 / speed beacons you have FIRST go onto your refineries, then you upgrade all your pumpjacks to Speed3 / Beacons. While ramping up Speed3 and Producctivity3 production, you'll need a hugely efficient furnace setup. And IMO, Steel makes a lot of sense for that. At the very, very endgame, you'll want to then transition into Productivity 3 / Speed Beacons on even Electric Furnaces all the way at this Iron Ore level.

And of course, to feed your Productivity3 / Speed refineries and pumpjacks, you'll need 100s of MW of energy. Fixing that issue needs a GW energy plant of some kind: over 17,000 Solar Panels (just a GW of energy. A lot of it would be lost in Accumulator storage at night).

So yeah. Post rocket, but before super-late game, I think this Steel Furnace study is useful. Still temporary because Speed3 / Productivity3 on Electric Furnaces is clearly superior. But a solution to tide us over between the points would be ideal.

---------

BTW: If you're going for the 20-million green circuits achievement, you'll need to have 27 blue belts of Green Circuits (fed by roughly 27 blue belts of Iron and 40.5 blue belts of copper) running for 5 hours. I would presume most mega-bases initial goal is to crush that 20-million green circuit achievement.

Because the footprint is the same size as the simple, tilable 11x70 design, I would guess the "real purpose" of the rainbow belt solution is to save on blue belts rather than on space.

It's actually a tile smaller when tiled because it shares a belt with the next strand.

Good point. That's still gets us a "practical" size of 700, which is still larger than 657. gheift is using some sort of magic...

The first tile is 11x70, all the rest are 10x70.

[siggboy]

That's still gets us a "practical" size of 700, which is still larger than 657. gheift is using some sort of magic...

"Practical" size is the key word here. For the 657 tiles of footprint that gheift advertises cannot be perfectly distributed into a square or rectangular area (without a remainder). So some of the space is going to be wasted.
Edit: OK, that's not really true, as I've found that 73*9 = 657, but I doubt his layout is a perfect rectangle with those dimensions.

I presume the "bounding box" of his design will still use the 700 tile area, making it equivalent to Blake's rainbow belt solution from a practical point of view.

Saving 43 tiles of space is not really that impressive if it's practically not usable for anything.

Gheift, can you post your solution please, we want to see it

[dragontamer5788]

That's still gets us a "practical" size of 700, which is still larger than 657. gheift is using some sort of magic...

"Practical" size is the key word here. For the 657 tiles of footprint that gheift advertises cannot be distributed into a square or rectangular area. So some of the space is going to be wasted.

Fair enough.

Nonetheless, I figured out a new pattern. I don't have the time to build a full-scale version, but here's a quickie solution that I'll map out later.

small.png (341.43 KiB) Viewed 36456 times

Dunno if I'm saving anything yet. But I'm gonna get two-rows of 140 x 7... but its 6x140 when tilable (because it shares output with the next lines) which would be 840 tilable tiles for 1-belt of output.

Which is more than 11x70. Crap...

[dragontamer5788]

Actually, this design tiles very well. I'll have to build it out in full-scale later.

Factorio 0.13.11 8_11_2016 11_38_39 AM.png (867.7 KiB) Viewed 36454 times

I only need a row of 35, now that I think of it. 12x70 for two belts of output == 840 footprint. Three belts would be doable with one more row. So 18x70 for 1260 footprint with this design in theory?

Someone check my math. Those are absurd numbers. I'm apparently halving the footprint. So either this is an awesome design or my math got all screwed up again.