Maximum express belt steel furnace throughput

[Cerbsen]

This setup of 54 steel or electric furnaces can achieve maximum express belt compression, with an iron or copper throughput of 14234 items in 8 minutes ( 1779 item per minute ).

[SHiRKiT]

I normally setup to fast transport belts and then combine into express one.

[quinor]

If I am right, theoretically, only 52 are needed.

[Alekthefirst]

52.4 unupgraded furnaces are needed to fill an express belt, but for symmetry you need 54

[Cerbsen]

52.4 unupgraded furnaces are needed to fill an express belt, but for symmetry you need 54

No, in therory you need 52,4 to bring the final belt to full compression, but 53 steel furnaces or 52 + 2 stone furnaces are not enough to guarantee 100% compression, there will be gaps in the final belt from time to time.

[Patric20878]

Can confirm OP with own tests that I ran earlier as well. As taken from my notes:

- Approximate Max Throughput of One-Sided Belts in Terms of Total Smelting Speed:
Normal Belt: 20 (3 Fast Inserters)
Fast Belt: 36 (5 Fast Inserters)
Express Belt: 53 (7 Fast Inserters)

Normal and express belt speed only roughly tested. Have confirmed though that 36 total smelting speed seems to be about the exact number needed for fast belts, which I now use in my 18 electric furnace design that has 2 speed modules each and beacons for pollution reducing. Which is equivalent to 36 steel furnaces, 2/3 of the 54 you got for express.

And @cerbsen: yep, that same gap issue is why I prefer my electric furnace setup in double rows of 18 total (2 speed-3's in each) outputting onto fast belts eventually instead of 27 onto express. In the 18-setup I can just use express belts to decompress it temporarily before the fast belts to get around the gap issue heh.

[nfcc08]

Always interested in seeing how the max of everything is achieved will have to try this out! Looks nice

[Cerbsen]

Normal and express belt speed only roughly tested. Have confirmed though that 36 total smelting speed seems to be about the exact number needed for fast belts, which I now use in my 18 electric furnace design that has 2 speed modules each and beacons for pollution reducing. Which is equivalent to 36 steel furnaces, 2/3 of the 54 you got for express.

And @cerbsen: yep, that same gap issue is why I prefer my electric furnace setup in double rows of 18 total (2 speed-3's in each) outputting onto fast belts eventually instead of 27 onto express. In the 18-setup I can just use express belts to decompress it temporarily before the fast belts to get around the gap issue heh.

You can find the maximum possible belt throughputs here: https://forums.factorio.com/wiki/inde ... ed_Results

For fast belts and iron smelting steel furnaces it would be 19,73 / 2 * ( 3,5 / 2 ) = 18 (17,3) furnaces per fast belt side to reach maximum belt compression or 19,73 / 2 * ( 3,5 / 4 ) = 9 (8,6) electric furnaces with 2 speed-3 modules.

[Patric20878]

You can find the maximum possible belt throughputs here: https://forums.factorio.com/wiki/inde ... ed_Results

For fast belts and iron smelting steel furnaces it would be 19,73 / 2 * ( 3,5 / 2 ) = 18 (17,3) furnaces per fast belt side to reach maximum belt compression or 19,73 / 2 * ( 3,5 / 4 ) = 9 (8,6) electric furnaces with 2 speed-3 modules.

My numbers already match that. That 18-electric-furnace setup of mine is 9 furnaces per side with 2 speed 3's each, same as what you said. And my chart is max belt throughput per side in TERMS of total smelting speed combined from all furnaces, not total number of furnaces or items/per second. The 36 I got for fast belts from testing: 9 furnaces per side, each with 2 speed 3's, is 4 smelting speed each. 9x4 = 36

The point of putting it in terms of total smelting speed instead of items/second is because the furnaces all have smelting speed listed on them and it greatly simplifies the math. For example, if I have 4 electric furnaces with 2 speed-3's each, 5 steel furnaces, and 3 stone furnaces outputting on one side of a fast belt. then the total smelting speed is 4x4 + 5x2 + 3x1 = 29. 29/36 = ~80% of max fast belt throughput. And unlike doing items/sec calculation, these are all whole numbers and I can easily the math for that mentally.

And actually, 8.6 is not accurate; 9 is probably not 100% accurate either, but it's easily proveable to be closer to the right value than 8.6 is, because I've done it already. Don't blindly believe in the wiki. I've proven multiple times in Tekkit Classic that even wiki can be inaccurate, as wiki is edited by people and very small and subtle mistakes can happen. If you are familiar with that and it's wiki, I've personally corrected at least 60% the pages there, and likely I'd do the same for Factorio too if I get the time to run such time consuming and thorough tests for all those values. The max throughput of fast belts is greater than 8.6 (9), as when I had 9 electric furnaces with 2 speed 3's outputting on a express belt that then compresses it into a fast belt, none of the electric furnaces started stacking stored iron plates. If the true max throughput were 8.6, the belt would start clogging with iron plates, inserters would not be able to pull iron from the last furnace, and so it would start stacking iron plates in its output slot. I'm almost certain that the items/second value on wiki was tested by outputting items directly on a fast belt. And guess what? That runs straight into the gap issue we were talking about above. If the belt is approaching full throughput and you don't output onto express belts, the items simply cannot decompress enough on the belt for additional items to get in, and then that's a matter of throughput being limited by compression. When I did the same thing, I also got a value VERY close value to 8.6

[Cerbsen]

I've seen that your numbers are correct

I just wanted to point out how to calculate belt usage for most cases instead of manual testing.

[Patric20878]

I'm aware yes. But I'm also pointing out that the method I use to calculate is much simpler to do. Give it a try, add smelting speeds on all furnaces outputting on a fast belt to get the total amount, divide that by 36, that's how maxed out the fast belt is.

And as said in the above post, the items/second value on the wiki is inaccurate. It was very likely done without taking into consideration the gap caused by not achieving full compression. A closer value than 1184 items/second assuming that 1184 matches 8.63 furnaces is 1235, from 9 furnaces.

And so if I know those two numbers, I can even get the items per second easily too. 29/36 is .805555. I can simply do .805555 x 1235 = 995 items/second, and that's probably a very close value to how much items per second is going on the belt, and I can get that number just from adding up smelting speeds on all furnaces on a side, no measuring needed at all. Simple isn't it?

And I should consult ssilk about this also, he posted those original items/second values up actually.

[ssilk]

The numbers in the wiki are >1.5 years old. Things may have changed a lot.




Btw, if anyone wants to repeat the measurements:

I built a timer first. I calculated how much time an item needs to run exactly one minute and made serveral tests, I think it was not really exactly one minute. But the function of that timer is: Item is placed on belt, one minute to go, item is placed into chest again. Let's call it "state engine".

The test-drive knows only two states: Preheating and running.

Preheating is done, when the time isn't started. This works like so:

- There are six rows (9 inserters each in one row), which place items on 3 express belts from both lanes.
- The three lanes are joined with fast splitters and
- the test drive starts. (state = running)
- The test drive itself is just some belts, curves or whatever form you want to test.
- Before and after the test-drive are no curves etc.
- After the test drive there are the "running-state"-inserters on one side of the belt (about 20 green inserters) and the "preheating-state" inserters on the other.

Preheating and running-states cannot happen at the same time! (see above: the time is a state engine).
Preheating is needed, because otherwise you do test how long it takes, until the first items run through. But you want to test the total throughput.


How this works in practice:
- In preheating state the items run through the test drive and the preheating inserters take them away at the end.
- When the time is started, the preheating inserters stops immediately and running-state inserters take over.
- When the timer runs out, the running-state inserters stops immediately and preheating inserters restart.

Now I have time to count the items. I measured it by taking the items out of the 20 boxes and filled them into two wagons, which made the counting really easy.

What I would change for this test-setup nowadays:
- Make the timer more exact. Perhaps we can use the train, which gives an more exact timer than belts. Or (much better!) use a mod. I think there is a mod with a timer? But found none.
- change the counting to something more elegant: Instead of wagons just an logistic network?
- Create a better state-engine: Preheating, Running, Counting, Reset.
- Enabling testing nifty stuff, like: Is there a difference, when placed one tile off (see https://forums.factorio.com/wiki/inde ... s/Priority), or having two inputs or outputs...
- Avoid any manual intervention after the test drive is set up. This was with my measurements always the biggest source of errors. For example: Faster reset, cause it tooks a while to empty the wagons and I emptied it sometimes by hand.
- Make a short timer, 10 seconds are good enough to test the test.
- Run several tests at once (blueprints).

[Patric20878]

Oh, cool, you replied. Was going to PM you heh. I'm so not used to forums that are actually active (Tekkit forums) xD. But yeah, I read that part on wiki where the tests need to be redone, don't worry.

Without pictures, it's hard to visualize what you're describing, and unfortunately, on your last section about needing a more exact timer, my setup isn't sensitive enough either to measure it to a precise number, though I do know that my setup has exceeded the 1184 items/sec throughput you posted, assuming cerb's math based on your value is correct.

In detail, the method I use is a simple one: decompress all output items on belt to minimize the bottlenecking caused by items not being able to be put on belts due to compression with an express belt on the electric furnace output belt. And take care that the furthest output belt from the miners that is also still between the electric furnaces is an express belt, because compression at near maxed out throughput is so sensitive that even a fast belt will completely screw it up and bottleneck it down to 8.6ish again.Haven't really tested longer lengths of express belts either, but I know that longer ones don't necessarily also work, so use that exact length. Anyways, what I did then was continually add more upgraded electric furnaces until the later furnaces' output slots started stacking up, which would mean that input is now exceeding output. By retesting again and again, but with finer measurements, finest being a stone furnace with just 1 smelting speed, I arrived at 36 smelting speed worth of max throughput for a fast belt (9 max speed upgraded electric furnaces). And then borrowing cerb's math, I concluded that your setup must not be accurate enough since his number of furnaces calculated from the 1184 number was ~8.63, which would be roughly 96% of the number of max upgraded furnaces I got to output on a fast belt continuously with none the furnaces starting to stack items. And so the math is, if it were 100% instead of 96%, the items per second should be ~1235, though this is all theory until I or someone else confirms this with vigorous ingame testing.

Here's pic of the setup: https://imgur.com/a/NLWtB#2
Had to test for fast belt throughput to figure out my ideal setup yeah. If you copy the setup exactly, you can see that once it gets going, not only does the electric furnaces not start stacking up, but the fast belt after the express one also appears completely compressed. But that I can't really just confirm, so all I can say is that I'm sure that it's at least 1235 based from the results, rather than equal to 1235.

I can also upload the test map for you instead if you prefer. And if you could upload your test setup as well, that'd be great too so I can see the method you're using. From what I understand, I'm heavily suspecting that your setup from 1.5 years ago didn't account for compression losses in output, as when I used fast belts between the electric furnaces, I got a number very similar to the one you posted as well. Maybe you can run my setup on the Factorio version you ran 1.5 years ago and see if the results still match?

[ssilk]

With v0.8 or 0.9 the belt behavior has been changed a bit. It's also quite possible, that the placement of the splitters plays a small role.
I need to look, if I can find that test map, but I think I posted it not at that time, cause I found it too ugly.

And in endeffect, what I can now say after that time: There is really no big difference in placing one more or less furnace (steam-engine, mine, whatever).

The important "schtuff" is:

- Are there obvious flaws in the throughput? I mean: A loss of more than 10% begins to hurts. Below that? Ask you: Is it worth the effort, or should I live with it, or (and that is eventually much better!):
- is there place enough to build the whole construction a second, third, fourth ... time, and split the incoming/outgoing into that many streams? Cause that multiplies suddenly the throughput!

With all I know, this last point is in most cases much better, that try to get the last percent throughput out of a complex setup. But on the other hand, that type of optimization is sometimes the most fun, when you found the real bottleneck.

[Cerbsen]

While testing belt compression i stumbled over strange belt behavior. If you are lucky, you can get a little over 1235 items per minute on a fast belt, if not you get less.

Here an example of what i mean with basic belts.

I build two basic belt lanes and six electric furnaces on side per belt. The plan was to fill a line chests at the end of each belt with the output of a full compression belt lane (six times 2 speed-3 modules) and an other with the output of six furnaces without modules. After 10-20 min. i wanted shut down the furnaces to compare the output of the six 12 times speed furnaces with the full compression belt and calculate the throughput at full compression.

But the placement of the merge point between the express and the basic belt changed the throughput of the belt.

The basic belt placement in the picture can transport ~370 items / min. per side. If you extend the basic belt to the yellow mark on the left, you only get ~354 items / minute. Merging at a red rectangle, you get ~345 items / min. I measured the rough amout by changing speed modules in the furnaces until i found the breakpoint between to much or to less throughput to provide full compression.

[Patric20878]

@ssilk: With how simple and effective my electric furnace setup is I get the best of both worlds And the % loss is actually quite noticeable. The decompression bottleneck and the belt merge point I mentioned and cerb measured and made a picture of would limit my setup to only be able to output 8 upgraded electric furnaces per side instead of 9, and that's an 11.11% loss already

@cerb: Yeah, that sounds exactly like what I was talking about with

And take care that the furthest output belt from the miners that is also still between the electric furnaces is an express belt, because compression at near maxed out throughput is so sensitive that even a fast belt will completely screw it up and bottleneck it down to 8.6ish again.Haven't really tested longer lengths of express belts either, but I know that longer ones don't necessarily also work, so use that exact length.

That 1235 number I calculated based off my setup already includes the optimal merge point so it achieves full compression from what I can see, so it won't be "higher if I'm lucky". Spent hours playing around with that setup to solve that problem and a few others until I saw none left, and the decompression and merge point were two of the hardest ones to figure out. Unless merge points are optimal, the furnaces don't output plates on optimal positions, ie right next to a plate, and that in turn causes decompressed output, so while there is space between plates (you get enough space by compression via express belt), there isn't enough for a plate to get in. No surprise I got hooked on Factorio thanks partly to belt mechanics and logic xD

Us three should work together on measuring up a more accurate max throughput on belts heh. This seems like deeper knowledge on belt mechanics that isn't well understood yet. And on a side note, I'm not sure if 2 partially full lanes merged by a splitter always achieves full compression if the input is enough either, so if that's true I don't really see any good way to measure max throughput of express belts when there's no faster belt to decompress the output belt. As I read, if the left belt merging into a splitter (in certain orientations) is fully compressed, then that would take priority and logically should always output fully compressed since it wouldn't even merge from right side, but if there's partial input on both belts, I'm not sure. Have any good ideas in mind to test express belts, ssilk/cerb?

[Cerbsen]

Have any good ideas in mind to test express belts, ssilk/cerb?

Modding in a super express belt or increase the speed of basic/fast belts.

With the two belt merge the higest throughput for express belts was about 1837 items / min. in the posted test setup, but it may be higher (~54 combined furnace speed) with another merge position.

[Patric20878]

Ah, that could work. Also test with my method, where you add furnaces until right before they start stacking up and belts become fully compressed. Have you tested for the max throughput of fast belts?

[ssilk]

Have any good ideas in mind to test express belts, ssilk/cerb?

My plan was to wait for better circuit network. Exact timers, easier measurements. And a scientific aspect of the different measurement methods would be to test, if they are really compareable.

[Cerbsen]

Ah, that could work. Also test with my method, where you add furnaces until right before they start stacking up and belts become fully compressed. Have you tested for the max throughput of fast belts?

Numbers for one belt side
* basic belts : 21,4 - 21,6
* fast belts : near perfect @ 36 combined smelting speed, 36,2 starts stacking
* express belts : minor/very slow stacking @ 54 smelting speed

just use 22 for basic belts as in early game you have no access to modules

express belts were done with double belt merging... you may download the testworld in my post above and rerun it with a single belt. The savegame contains 3 seperate furnace setups (one per belt type), steam engines to provide up to 28 MW power for furnaces and way to many electric miners with about 1 million iron each.