UPS optimized smelting - belts only

[Belter]

I'm working on UPS optimization for my factories to run better on my potato computer.

Among other things focusing on smelting - using a fully beaconed smelting line to produce a full blue line.

After some trial and error I've found this layout:
- UPS optimized: AFAIK the less smelters/inserters I have, the better
- power/pollution is not an issue, end game phase
- must be tilable, 100% vanilla, belts only
- the last smelter on the righ is to fill the remaining gaps, so I've added buffering. Less could do as well
- interesting that doing a mixed SM3+PM3 smelter mods would be "safer" than a 2xPM3 setup (needs more ore but 8.91 vs 8.96 smelters)
- layout could be compressed 1 title in height, but would need more belts & I like the lights this way
- a bit offended that it ends up at 9 smelters, why not 8???

Smelting-UPSoptimizedMaxBeacon-clean.png (1.86 MiB) Viewed 7968 times

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

Copy blueprint

Smelting-UPSoptimizedMaxBeacon-PMPM.png (68.1 KiB) Viewed 7973 times

calc

Smelting-UPSoptimizedMaxBeacon-PMSM.png (68.19 KiB) Viewed 7973 times

calc

A SM3 only setup needs exactly 9 smelters -> and a full belt of ore of course.

Comments welcome before I refactor my bases!

[mrvn]

Have you tried using circuit control on the copper plate inserters so they activate on a timer whenever 12 plates are ready for pickup? I think that saves some UPS.

[DaveMcW]

UPS optimized

I like the lights

Contradiction!

[Belter]

Have you tried using circuit control on the copper plate inserters so they activate on a timer whenever 12 plates are ready for pickup? I think that saves some UPS.

Yep. I see mostly 2 plates (sometimes 1 or 3 due to prod mod) put on the belts so this defentely makes sense.

The first 8 inserters are working like ~continuously.

Game should have a minial stack size setting for inserters...

Thank you for the comment!

[Belter]

UPS optimized

I like the lights

Contradiction!

I assume lights have effect on FPS and no/minimal on UPS. Are they so bad for UPS??

Edit: I did a test, no/tiny effect. I have 38k, but not circuit controller lamps. By deleting all I see no difference, still around 40fps. The 192MW they consume during the night is also not a big deal compared to the 24GW need.

The command to delete all lamps I've found here: viewtopic.php?t=61054

[wobbycarly]

UPS optimized

I like the lights

Contradiction!

If they were wired to a constant combinator to keep them on all the time, would that save UPS?

[mmmPI]

If they were wired to a constant combinator to keep them on all the time, would that save UPS?

From the added link in the edit after you posted it would interestingly seem like wiring lamps to circuit could be more costly than not wiring them at all.

Having lamps in the first place is more costly than having none in theory because their purpose is not production but it may not be enough to be visible in practice depending how precise you want to do the testings. At least they are not blinking funny colors in the trade-off "look-good"vs"UPS efficient"

If deleting 38K lamps doesn't yield any visible result, you can safely add 200 more, but this is on a particular computer with a particular factory. If you have a very inefficient nuclear reactor ups-wise, 192MW could be costly in UPS and thus would prefer not having the lamps to use that energy for something else. ( 40fps still but more spm ! ).

24 GW is a lot of power for a potato computer anyway, I have one hehe and I know i would use solar in this case .

Game should have a minial stack size setting for inserters...

There may be something you can do with the inserter grabbing the copper ore, if you manage to synchronize them, then you can activate both at the same time when you know the furnace will have produce enough to make better use of their stack size.

Or with other word, if you time it right, you can have a cycle of say 5.2 second, and activate both inserters 0.3 second every 5.2 second. those are made up illustrative numbers.

Even with productivity modules the input frequency is the same. The output frequency can be the same too, you just have to ouput say 10 or 11 when there is an extra plate due to the module. Then you input and ouput at the same time for 1 swing every interval.

Input stack size=> 10 output stack size => 11 or 12 to account for productivity.

I would make a self reseting clock that has a period of a little more than the time it takes for a furnace to smelt 10 plate say count from 1 to 256 and when the clock is under 15 I would enable both inserter. Since you need 8.96 furnace, it's ok if the clock is not perfect, you can loose a little bit of time. It would "slow" a little the 8 first furnace, so that all 9 function at the same time.

Or you can have only 8 synced, and the last one filling up the gap to make it easier to find a timing.

But i'm not sure it would save any UPS and at which scale it would start to have an impact



EDIT : I had a second read on that and noticed the "belt only" in the "circuit free" nevermind the self reseting clock.
This is an attempt at minimalist setup :

Belt-clock synchronisation

stacksize12.png (818.66 KiB) Viewed 7831 times

It's still using circuit but minimum logic, I tried various belt combinaison for the loop, this one with the plate on the outer lane makes inserter output 10 or 11 plate at a time every lap and not quite 12 as i tried.

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

Copy blueprint

[Belter]

24 GW is a lot of power for a potato computer anyway, I have one hehe and I know i would use solar in this case .

Yes, I have solar only power - also to be optimized (robotborts still there). But looks really awesome

It's still using circuit but minimum logic, I tried various belt combinaison for the loop, this one with the plate on the outer lane makes inserter output 10 or 11 plate at a time every lap and not quite 12 as i tried...

This is really impressive and like it a lot! It take me some time to realize that I have to put somehing on the belt as its not coming in the blueprint...

Edit: I missed that one of the belts are blue, speed is abolustely OK w/the blueprint!

Edit: I keep it here just as a fun measurement, per mmmPI's replies below speed can be fine-tuned by mixing belt types:

SimpleCircuit-Tests.png (679.36 KiB) Viewed 7770 times

4-long (circle), outer lane: ~20 pulse/min (->gaps)
6-long, inner lane: ~22 pulse/min (->gives me a full blue belt, grabs 8..11 plates, yes its less good looking though...)
4-long (cirlce), inner lane: ~50/min (->too quick)


Thank you for your replies, really interesting points, woah! Seeing the inserters doing much less swings is really cool!

[mmmPI]

Timing is a bit too slow so to be adjusted imho

But i thought i did the adjustment !

1)4-long (circle), outer lane: ~20 pulse/min (->gaps)
2)6-long, inner lane: ~22 pulse/min (->gives me a full blue belt, grabs 8..11 plates, yes its less good looking though...)
3)4-long (cirlce), inner lane: ~50/min (->too quick)

According to my previous measurement from this post : viewtopic.php?f=18&t=100292

there are 60 ticks in 1 second.

1) => 147.5 ticks per lap (more than 2 second)
2) => 117 ticks per lap (a bit less than 2 second)
3) => 53 ticks per lap (a bit less than 1 second)

You did found the value 117 ticks to be the best.

In the blue print i used it's meant to be 3 yellow belt and 1 blue belt. And the plate on the outer/long circle.

This should gives 122.9166 ticks per lap or very close to the 117 you found. I did the adjustment with a stack size of 12 for the filter inserters.

I tried 3 yellow 1 red and thought it was too slow so i used 3 yellow 1 blue. Maybe i didn't tested long enough and some buffered plate fooled me.

Well that's just details anyway, you got it, you could easily adjust this for steel or stone brick

Using the same item in the loop as the item outputed allows to use "set filter" on the output inserter, but you can also use any item and stack inserter wired to "enable/disable" IF Anything >0 for the output instead of the stack filter inserter. I'm not sure which one is better for UPS. Or if it helps at all x).

Looking at the blueprint again i really like the last furnace side loading with 2 inserters, easy and efficient.

[disentius]

Clocking inserters saves ups in big megabases (>10k SPM).
Here is How:
https://www.youtube.com/watch?v=kHvqxKL ... t3&index=4

[mrvn]

Would be nice to have an inserter that places an item into the belt clock after blueprinting.

But anyway, clocking the inserter means the design has circuits so it doesn't really belong in this forum anymore then.

[mmmPI]

[Moderated by Koub : Off topic]

[Belter]

In the blue print i used it's meant to be 3 yellow belt and 1 blue belt. And the plate on the outer/long circle.

I have to apoligize! I was playing around with your blueprint, re-created it & missed that 1 of the belts was blue! Works like charm!

Using the same item in the loop as the item outputed allows to use "set filter" on the output inserter, but you can also use any item and stack inserter wired to "enable/disable" IF Anything >0 for the output instead of the stack filter inserter. I'm not sure which one is better for UPS. Or if it helps at all x).

That'd be a good exercie to benchmark, both works, also on both "pulse" and "hold" modes. As my factory is mass producing stack inserters, I prefer those over filter inserters:

StackInserterOnOffSetting.png (38.2 KiB) Viewed 7544 times

Looking at the blueprint again i really like the last furnace side loading with 2 inserters, easy and efficient.

TY! Also helps to see on map view which diretion the block is.

The only issues is that I cannot put iron plate and steel next to each other, as this output in the middle. Would be nice only to save the last column of beacons (both resource and space) but who cares in late game... In case of top input, bottom output you can flip and connect, then you need more belts for the output.

Actually you made me consider no circuit -> minimal circuit for my base!

[mmmPI]

That'd be a good exercie to benchmark, both works, also on both "pulse" and "hold" modes. As my factory is mass producing stack inserters, I prefer those over filter inserters:

Yeah makes sense the stack filter inserter are often not a priority to automate the already rare case when they are automated at all .

I tried several setup i'm glad i didn't made a mistake when sending the "good" one it happens sometimes

The white stack inserter cannot have 2 filter set via circuit at the same time i think.

I wondered when testing with 1 loop belt wired on 2 side to each inserter. Like left side was wired to the input inserter and made with yellow belt, and the right side was connected to the output inserter and made with red belt. Then on this loop i put 2 item attempting to have them on opposite diamater, one copper plate, one copper ore. It was an attempt an clocking both input and ouput inserter with a single loop. It sort of work but didn't look clean.

Knowing that input inserter need to chase item on belt they couldn't in this configuration be activated by a pulse signal. It is too short of a duration. So the 2 yellow belt were set to "read hold" and i'm not sure for the reds if it was "1 pulse" and "1 nothing" or 2 "read hold".

There could be several slight variations like the 6 belt loop on the inner side, or the 4 belt loop on the outter side. The version with 6 belt having a straight portion and being 3 tile wide it makes it easy to place an item exactly at the opposite of one another.

Other little trick that exist if you want to place those items at the exact same time is to use an inserter only powered by a single solar pannel to drop the item on the loop. If you do this overnight, then the next morning all the inserters using this method will activate at the same time.

Using the 6 belt version it's possible to use such inserter from the top side for copper clocking loop and from the bottom side for iron clocking loop. This way you mitigate the power spike. (and potentially the ups fluctuations). ( the 4 belt version is not symetrical so it's not as good for this purpose ).

The only issues is that I cannot put iron plate and steel next to each other, as this output in the middle. Would be nice only to save the last column of beacons (both resource and space) but who cares in late game... In case of top input, bottom output you can flip and connect, then you need more belts for the output.

Actually you made me consider no circuit -> minimal circuit for my base!

i'm not understanding that part about iron and steel. I see the first furnace of the array having a 1 tile vertical spacing on the left side between the beacon. I would paste the ore=>plate on the left side, and paste another one identical copy just on the right side reusing the last row of beacons. And having the belt full of plate using the vertical gap to become the input belt for the second copy that would make the steel.

then you need more belts for the output of the steel, same conclusion ( if steel furnaces uses prod modules )

If you were to move the 1rst input inserter 1 tile on the right side, you could probably use 2 time the blueprint back to back for copper with a long array of underground belt bringing the ore for the 2nd blueprint, and then at the same location it would be a long array of underground belt outputing the plate from the 1rst blueprint. ( just 1 tile above the existing input belt + little adaptation for the last furnace).

I believe it's ok to post on this forum because the first post was clearly a furnace array circuit-less, then there was a suggestion from a user to clock inserter with circuit, then the same user seem to suggest the thread has gone off topic. Then my comment about that user is moderated. But not the topic itself (hopefully).I think it means it's ok to discuss about the original creation and adding circuits to it. ( and i'm sorry i wasn't very respectful toward another user whose initial suggestion was useful).

The video linked by @disentius is one that i have used to learn those tricks The scale is important for the trade off between the control mechanism's fix cost and the amount of computation it saves on 1, 10, 100000 inserters that's what i learned.

[Belter]

I wondered when testing with 1 loop belt wired on 2 side to each inserter. Like left side was wired to the input inserter and made with yellow belt, and the right side was connected to the output inserter and made with red belt. Then on this loop i put 2 item attempting to have them on opposite diameter, one copper plate, one copper ore. It was an attempt an clocking both input and output inserter with a single loop. It sort of work but didn't look clean.

Here the input inserters are working pretty nice and UPS efficient grabbing usually 12 items -> I'd not add logic here to control that. Interesting idea still, funny to see the 2 items going round!

The only issues is that I cannot put iron plate and steel next to each other[...]

i'm not understanding that part about iron and steel[...]

You're right. I had my 8-beacon setup in my mind where output and input were up and down, so I was able to mirror and put them after each other and have the same vertical alignment, OCD compatible. Does not apply here as I to have an output that way in the 12-beacon setup the pattern should be extended one tile horizontally - does not worth. I’d like to have a minimized 12-beacon setup now and get rid of my loved lamps between the beacons, TBD.

I believe it's ok to post on this forum because the first post was clearly a furnace array circuit-less, then there was a suggestion from a user to clock inserter with circuit, then the same user seem to suggest the thread has gone off topic. Then my comment about that user is moderated. But not the topic itself (hopefully). I think it means it's ok to discuss about the original creation and adding circuits to it. (and i'm sorry i wasn't very respectful toward another user whose initial suggestion was useful).

I'll open another thread in the circuit topic as I'm sorry that your post got moderated out. +have some other questions if there is interest

[Belter]

The new thread is here: viewtopic.php?f=193&t=100763