Factorio Forums

www.factorio.com
https://forums.factorio.com/

UPS optimized Steel smelting to 1 blue belt

https://forums.factorio.com/viewtopic.php?f=193&t=101147

Page 2 of 2

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Sat Jan 15, 2022 6:26 pm
by BlueTemplar
Heh, I was wondering how long it would take for the direct miner insertion to pop up !

BTW, UPS-wise, are you sure that you want to use prod modules rather than speed modules in those furnaces ?

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Sat Jan 15, 2022 7:23 pm
by disentius
here is the test with all three.
- 25000 ticks
- 10 runs
(dave's save had one weird much higher result, no clue how that happened. I did not do anything on the test computer while running the test)
2022-01-15 20_17_39-bm_ results.csv - LibreOffice Calc.png
2022-01-15 20_17_39-bm_ results.csv - LibreOffice Calc.png (118.09 KiB) Viewed 4471 times
testfiles attached.

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Sat Jan 15, 2022 7:28 pm
by disentius
@BlueTemplar:
yes. these are for real world late game use. also, I believe using only speed will not be as fast due to the increased use of iron, leading to more swings and more active % per steel bar.
But if you have a design, I'll be happy to test it.

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Tue Jan 18, 2022 3:30 pm
by causa-sui
Agree, I would expect prod modules in the furnaces to perform better than speed. But I would love to be proven wrong about this.

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Sun Jan 23, 2022 7:57 pm
by Belter
I've optmized further the di layout: timed the steel (848) and iron (206) inserters, removed a row of electric poles and cleaned up the output belt.

Next step is to measure the different versions above, and play a bit more... hope I can do that the next weekend! I'd like to understand the impact of:
- separate/merged circuits (guess: small or none)
- if timing perfeclty (no slipping items on the belts) gives any UPS (guess: nope)

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Sun Jan 23, 2022 9:21 pm
by Belter
causa-sui wrote: ↑
Tue Jan 18, 2022 3:30 pm
 Agree, I would expect prod modules in the furnaces to perform better than speed. But I would love to be proven wrong about this.
Edit: I just calculated the effect of the modules, did not do any benchmarks

Actually, in high beacon setups 1-1 is the best, but the difference is really small. And the fact that the input resource need is smaller makes Prod modules usage an easy decision for me UPS wise... producing 20% less iron plate should be an easy win.

See my table below, hope numbers are correct.

I did such calcs for assembler machines a while ago, that was a bit more surprising, see Speed vs Productivity modules: The 12 bacon suprise

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Mon Jan 24, 2022 4:29 pm
by disentius
This one is slower than your previous one.
2022-01-24 16_18_19-Window.png
2022-01-24 16_18_19-Window.png (46.59 KiB) Viewed 4305 times

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Tue Mar 08, 2022 1:25 pm
by BlueTemplar
disentius wrote: ↑
Sat Jan 15, 2022 7:28 pm
 @BlueTemplar:
yes. these are for real world late game use. also, I believe using only speed will not be as fast due to the increased use of iron, leading to more swings and more active % per steel bar.
But if you have a design, I'll be happy to test it.
Well, with direct insertion swings aren't an issue, and since this is about megabases, ore shortage shouldn't be one either ?

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Tue Mar 08, 2022 5:30 pm
by disentius
Just did a calculator check, you need 7 more furnaces per blue belt....:) no go.

12 beacons speed only:
https://factoriolab.github.io/list?z=eJ ... MGAIwtCDs_

12 beacons prod in furnaces:
https://factoriolab.github.io/list?z=eJ ... MGAIv6CDg_

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Wed Mar 09, 2022 12:37 am
by DaveMcW
Speed in the steel furnaces is clearly bad.

Speed in the iron furnaces might make sense. Especially if you are direct inserting from mining drills.

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Mon Mar 21, 2022 10:29 am
by fallenghostplayer
I am glad I found this thread before adding a 12 beacon steel build to my factory! I don't know if this is still a thread in progress, but if that is ok, I hope to add a bit to it.

Based on the concepts elaborated in the links below, I have come to changes in design:
-Mechanics of Transport lines by Smurphy1 (https://www.reddit.com/r/technicalfacto ... ne_splits/)
-UPS efficient double sided belts by Stevetrov (https://www.reddit.com/r/technicalfacto ... e_item_on/)
-What I learned by interacting with the above folks for my previous red factory (https://www.reddit.com/r/technicalfacto ... d_science/)

For testing, I use my own batch file that does the exact same but less fancy. I have used 50 alternating tries of each (so map_1, map_2, map_3, then map_1, etc, 50x), given there is alot of variability for the same map, even if nothing seems to be running in the background. Anything above 3 standard deviations is excluded as it is much more likely to be a one time ping for the antivirus/windows/etc than actual meaningful and repeatable UPS effects. Typically this means removal of 0 to 2 samples (ex: that big spike in the January 15 v5 results). 50000 ticks used.

Also, two half-belt inputs would be best for each lane, or even some variation on the number of lanes (why 6?) would be ideal, but it is unclear how the mod's infinite source/sink compare to in-game miners in terms of UPS per item "produced"/"consumed". The effect here would be to bias the results towards more or less lanes in comparison to half lanes produced by miners/trains/etc, so I did not investigate those avenues with the mod's sources & sinks.
steel_results_21_march.png
steel_results_21_march.png (8.15 KiB) Viewed 3867 times
10 beacon: Using disentius's v3 design as a starter, and with the techniques above, I got a bit more than 2% improvement
11 beacon: With an alternate layout using a few less beacons that DaveMcW's v5, and the techniques above, the 11 beacon build is very close to the 10 beacon build (less than 0.7%)
"Real case": To ensure a compressed output at a given output rate (here 45/s), it is usual to have slight overproduction to ensure back pressure exists on the client factory, which typically is good for UPS. With this idea, the furnaces are clocked more aggressively (843/1 for 10b and 783/1 for 11b) and have one factory outputting to a stub. Typically this also means you can use only the required assemblers/furnaces rather than underclocking them. Since the furnace will be full and an inserter waiting on a stopped transport line is sleeping according to the game engine, UPS usage is low, and ensures that the output rate is always maintained. The effect of using this idea is negligible in both 10 and 11 beacon builds.

I hope I didn't miss anything!

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Mon Mar 21, 2022 1:28 pm
by disentius
Awesome:) More views/testers more fun reproducible results. I recently found the smurphy post, and missed the stevetrov one. Study time, and I'll repeat your runs this week.

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Fri Mar 25, 2022 12:54 pm
by fallenghostplayer
For comparison, I've tested a 12 beacon design, specifically the one used by Smurphy1 in his UPS record megabase and I must say it compares rather well. Sure, its 2.7% slower than the current best 10 beacon build, but it definitely does not support the idea that 12 beacon builds are terribly inefficient.
steel results 25 march.png
steel results 25 march.png (9.21 KiB) Viewed 3760 times

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Sat Sep 02, 2023 3:04 pm
by arthurdent86
I tried to design a 8 beacon design that would be as ups efficient as possible. In my speedtests it was about as fast as disentius' design (which I modified slightly to get rid of the splitter), which is weird because that one uses 10 beacons. In both cases I used a map that produces 100 blue belts of steel. It would be really cool if someone could try to confirm my result since I've never done speedtests before and I might well have made some mistakes in testing this. I'll attach both maps I used (the map with disentius' design is called Speedtest_imported.zip)

edit: And can someone explain to me why my saves are so much larger than everybody else's?

This is my blueprint:


And this is the design by disentius I used (I made it kinda ugly, but it works):

Re: UPS optimized Steel smelting to 1 blue belt

Posted: Mon Sep 04, 2023 5:52 pm
by disentius
Hi Dent, welcome:)

Tips:
Use an empty editor map, with 2 mods:
- RegionCloner
- DeleteEmptyChunks (That is how to keep the map small.)
- In editor mode, you can set everything to map tiles, and do all research.
- You can use the express loader and infinity chest for output too, no need for inserters. (see maps enclosed)
- I use a script from velit to automate the tests, and run a local instance of the latest factorio experimental. (script is here: https://github.com/velit/Factorio-Benchmark-Powershell)

I see that I did my original test with Editor Extensions. It is not needed, so I ditched it.
Here are test of your setup, my original V3, and an smurphy1-advice optimized V4
test maps and empty maps attached. all maps have 50 blue belts of steel produced(ref: you would need 36.150 steel belts for 10K science per minute, including military). tests ran for 50000 ticks, each map ran 5 times.
Yours does as well as my V3:)
2023-09-04 19_49_28-steel smelting results.png
2023-09-04 19_49_28-steel smelting results.png (68.29 KiB) Viewed 1494 times
All times are UTC
Page 2 of 2

Powered by phpBB® Forum Software © phpBB Limited