Page 1 of 3

Challenge: maximum net energy from a blue belt of coal

Posted: Sun Jun 09, 2019 1:40 pm
by vanatteveldt
Everyone can run a belt of coal to X boilers to produce energy. But what is the maximum net energy you can get from a blue belt of coal?

Rules:
- creative mode: you get one matter source of coal, as many water fluid sources as you want
- goal: maximize sustained net electricity output, e.g. production minus consumption.
- scoring is consumption by passive void on 10m electricity view (with the assumption that it says the same, i.e. if I run it for X hours it should still give that number)
- you're allowed to use any number of buildings, modules, etc. You can use other items (e.g. heavy oil) as a catalyst, but cannot require a supply.
- No cheating :-) (e.g. using solar panels, other creative mode items, ...)

Baseline:
- 45 coal/s * 4MJ/coal=180MW, which can be consumed by 100 boilers, which with yellow inserters consume about .225MW, so net energy is 197.78MW

I'm sure we can do better than that, right? :)

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Sun Jun 09, 2019 4:24 pm
by vanatteveldt
First entry: 669 MW net production (681 calculated, maybe it needs more time to get there...)

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Mon Jun 10, 2019 9:24 am
by DaveMcW
704 MW net production without using any efficiency modules.

coal-power.jpg
coal-power.jpg (342.98 KiB) Viewed 845 times

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Mon Jun 10, 2019 7:04 pm
by Koub
This allowed me to generate a tiny bit more. I'm sure someone will do better, but at least I held the 1st place for a short time :)

[Edit] Not as good as DaveMcW's design : used steam source instead of producing my own.



705 MW net power generated :)
screenshot

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Mon Jun 10, 2019 9:50 pm
by DaveMcW
I don't think cheating for free steam is allowed. :P

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Mon Jun 10, 2019 10:02 pm
by Koub
The rules said I could have any number of fluid sources as I wanted. I assumed this included steam. I'll adapt my design when I have time, and try to see if I can do better.

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Mon Jun 10, 2019 11:24 pm
by patrick12222010
Can I use coal liquefaction? Then, I will convert coal to rocket fuel with prod modules. This will now power the boilers. See this post viewtopic.php?f=208&t=70916. I was able to achieve actual game Overunity of 2.09.

Edit: Just found out that coal liquefaction was already used. :)

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Mon Jun 10, 2019 11:31 pm
by vanatteveldt
Liquefaction is fine (not sure how else to get more power?)

Water fluid is ok, steam is not (op specifically mentions water fluid sources). But steam should be a small fractions of energy costs, main gains will probably be optimising modules and beacons..

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Tue Jun 11, 2019 12:04 am
by DaveMcW
FactorioBot wrote:
Tue Feb 26, 2019 5:13 pm
Relative fuel value of rocket fuel decreased by 10.6%, relative value of solid fuel decreased by 4%.
Rocket fuel was nerfed in Factorio 0.17, it is no longer profitable to convert solid fuel to rocket fuel.

You might come out ahead if you play version 0.16, even with less coal on a blue belt!

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Tue Jun 11, 2019 6:09 am
by vanatteveldt
DaveMcW wrote:
Tue Jun 11, 2019 12:04 am
FactorioBot wrote:
Tue Feb 26, 2019 5:13 pm
Relative fuel value of rocket fuel decreased by 10.6%, relative value of solid fuel decreased by 4%.
Rocket fuel was nerfed in Factorio 0.17, it is no longer profitable to convert solid fuel to rocket fuel.

You might come out ahead if you play version 0.16, even with less coal on a blue belt!
According to the wiki, solid is 12MJ. 10 solid makes 1.4 rocket fuel (with 4 prod modules) or 140MJ, so a gain of 20MJ. Of course you need to pay the assembler and presumably beacon out of that. Don't have the game here, what is the consumption formula for a boosted assembler? Our am I missing something obvious (again)?

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Tue Jun 11, 2019 11:23 am
by eradicator
Gosh. Can i say i hate you for making this thread?

Here's what you did to me:
  1. Hey there's an interesting challenge!
  2. Let's try with efficiency modules only
  3. 384MW without bacons
  4. Hm..what do the other people...700MW?! My f***
  5. I don't want to do the math.
  6. I do the math.
  7. I fall into despair because once again prod/speed bacon wins. Can there not be a single f***ing thing in this game that's better without bacons? I hate bacon sandwich.
vanatteveldt wrote:
Sun Jun 09, 2019 4:24 pm
First entry: 669 MW net production (681 calculated, maybe it needs more time to get there...)
No proof? → Didn't happen.
vanatteveldt wrote:
Tue Jun 11, 2019 6:09 am
According to the wiki, solid is 12MJ. 10 solid makes 1.4 rocket fuel (with 4 prod modules) or 140MJ, so a gain of 20MJ. Of course you need to pay the assembler and presumably bacon out of that. Don't have the game here, what is the consumption formula for a boosted assembler? Our am I missing something obvious (again)?
You're missing that rocket fuel takes 30 seconds to make. Which makes the cost skyrocket (pun intended). After numerous trials i found one *theoretical* configuration that might be able to gain 4% energy from conversion to rocket fuel, but i'm not sure if it can actually be built. Also as bacon sharing significantly reduces cost of each bacon you also need to build large enough to get a sufficiently low "bacon per machine" ratio. And one belt of coal might not be large enough.

Here's "the math". First define the calculation function:

Code: Select all

/c
local p = game.player
function get_power_per_cycle(entity_with_recipe,bacon_consumption)
  
    bacon_consumption = tonumber(bacon_consumption) or 0
    
    local this = entity_with_recipe
    local prot = this.prototype
    local recipe = this.get_recipe()
    if not recipe then p.print('<'..this.name..'> has no recipe set.') return end
    
    local bonus_speed        = ((this.effects or {}) .speed        or {}).bonus or 0
    local bonus_productivity = ((this.effects or {}) .productivity or {}).bonus or 0
    local bonus_consumption  = ((this.effects or {}) .consumption  or {}).bonus or 0
    local crafting_speed = this.prototype.crafting_speed
    local energy         = recipe.energy
    local consumption    = this.prototype.energy_usage
    local drain          = this.prototype.electric_energy_source_prototype.drain or 0
    

    local cycle_length     = energy * (1/(crafting_speed + bonus_speed)) * (1/(1+bonus_productivity))
    local power_per_second = 60 * consumption * (1+bonus_consumption) + bacon_consumption + drain
    
    p.print(string.format('<%s> consumes %.2fMJ per cycle.',this.name,cycle_length*power_per_second/1000^2))
    
    end
    
Then hover your mouse over any crafting machine and use this command. You'll need to input the number of *average bacons per machine*. I.e. if you have 10 assemblers enhanced by 16 bacons the number will be "16/10".

Code: Select all

/c
get_power_per_cycle(game.player.selected,480*1000*NUMBER_OF_BACONS_PER_ASSEMBLER)

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Tue Jun 11, 2019 7:03 pm
by Koub
I think rocket fuel won't work :
You can turn the 4028 solid fuel you get every minute from a coal belt into 564 rocket fuel.
Sure, it's a net gain of ... 8 GJ every minute, so 134 MW.
But for that, you'll need 137 MW of assembling machine 3, plus 24 MW for the 50ish beacons with speed modules.
If I trust the previous studies about how AM3 with 4 prod modules + speed moduled beacon sandwich is the optimal layout, then there's no way it can out net gain.

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Tue Jun 11, 2019 7:56 pm
by eradicator
Koub wrote:
Tue Jun 11, 2019 7:03 pm
If I trust the previous studies about how AM3 with 4 prod modules + speed moduled beacon sandwich is the optimal layout, then there's no way it can out net gain.
The problem is scale. Any sandwich of K rows of assemblers needs K+1 rows of beacons. The larger K is the less relevant is the "+1" fraction. Therefore if you build sufficiently large at some point you can make a tiny tiny surplus. If my math is right it works if you have on average less than 1.1 beacons per rocket fuel assembler.

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Tue Jun 11, 2019 9:02 pm
by Koub
Let's consider a single AM3 with 4 prod modules, sandwiched between 2x4 beacons with speed modules.
It will turn 110 Solid fuel (total 1320 MJ) into 15.4 rocket fuels (1540 MJ) on average every minute.
Dividing this will give us the output in MW, so 22 MW of solid fuel are converted into 25.666... MW of rocket fuel. So by doing so, you get 3.666... MW "for free".

To do so, the assembling machine 3 has 880% energy consumtion, so it consumes 9.8x375 kW = 3.675 MW, which is > to 3.666. So got your "free" energy at the cost of more energy. Lucky you, you didn't break the 1st thermodynamic law this time :mrgreen: (even though I don't care to break it in a game).

Even without considering the beacon's consumption, you need more energy to convert your solid fuel into rocket fuel than you get from doing so. For pure efficiency standpoint, burning solid fuel is better (if my math is correct, of course).

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Tue Jun 11, 2019 11:31 pm
by DaveMcW
Koub wrote:
Tue Jun 11, 2019 9:02 pm
Even without considering the beacon's consumption, you need more energy to convert your solid fuel into rocket fuel than you get from doing so.
I agree, at least for 8 beacons. The rocket fuel recipe gives +20 MJ. The cost of producing one rocket fuel recipe with 8x beacons is:
0.375 * (1 + 8.8) * 30 / 5.5 = 20.05 MJ

Image

With 9x beacons: 0.375 * (1 + 9.5) * 30 / 6.125 = 19.29 MJ
With 10x beacons: 0.375 * (1 + 10.2) * 30 / 6.75 = 18.67 MJ
With 11x beacons: 0.375 * (1 + 10.9) * 30 / 7.375 = 18.15 MJ
With 12x beacons: 0.375 * (1 + 11.6) * 30 / 8 = 17.72 MJ

So we did get some net energy with more beacons!

Unfortunately now we have to pay for the beacons, and that is impossible with an energy budget less than 2.3 MJ.

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Wed Jun 12, 2019 7:05 am
by quinor
Hmm, would adding efficiency 3 modules in beacons help?

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Wed Jun 12, 2019 7:13 am
by eradicator
Koub wrote:
Tue Jun 11, 2019 9:02 pm
Let's consider a single AM3 with 4 prod modules, sandwiched between 2x4 beacons with speed modules.
Maybe i'm being unclear. You example has an average of 8 beacons per assembler...

What i said is that this kind of "small scale" example CAN NEVER WORK. If you instead sandwich i.e. 10 rows of assemblers between 11 rows of beacons with same-length rows, you have 1.1 beacons per assembler. I said 1.1 is still too large so maybe if we go to a scale of 20/21 rows with an average of 1.05 beacons per assembler it starts working.

Have you tried using the function i posted above which does all the calculations for you? For the 8/1 example i get a cost of 16.94MJ per converted rocket furl for zero-energy beacons. Which is 3.06MJ less than is gained. With one beacon this goes up to 19.15MJ, still less than the 20MJ gained.
quinor wrote:
Wed Jun 12, 2019 7:05 am
Hmm, would adding efficiency 3 modules in beacons help?
So far every configuration i tried was worse.

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Wed Jun 12, 2019 9:29 am
by Koub
eradicator wrote:
Wed Jun 12, 2019 7:13 am
Koub wrote:
Tue Jun 11, 2019 9:02 pm
Let's consider a single AM3 with 4 prod modules, sandwiched between 2x4 beacons with speed modules.
Maybe i'm being unclear. You example has an average of 8 beacons per assembler...

What i said is that this kind of "small scale" example CAN NEVER WORK. If you instead sandwich i.e. 10 rows of assemblers between 11 rows of beacons with same-length rows, you have 1.1 beacons per assembler. I said 1.1 is still too large so maybe if we go to a scale of 20/21 rows with an average of 1.05 beacons per assembler it starts working.

Have you tried using the function i posted above which does all the calculations for you? For the 8/1 example i get a cost of 16.94MJ per converted rocket furl for zero-energy beacons. Which is 3.06MJ less than is gained. With one beacon this goes up to 19.15MJ, still less than the 20MJ gained.
No offense, but there was no use even trying : if you read thoroughly my post, you'll see that I didn't even account for the consumption of the beacons. I just took the AM3's boosted consumption and compared it with the gain due to productivity. Even with free beacon effect consuming 0 power, as long as you stack them with the classical layout, you'd still be losing some energy just from the assembling machines.

Dave McW did a complement to my post here : viewtopic.php?p=436070#p436070
Long story short, one would need to add more beacons per AM3, and this would still be too expensive in power with the vanilla beacons.
eradicator wrote:
Wed Jun 12, 2019 7:13 am
quinor wrote:
Wed Jun 12, 2019 7:05 am
Hmm, would adding efficiency 3 modules in beacons help?
So far every configuration i tried was worse.
Same for me. And same with all the tests that have been made so far ever since the very first one. This is a result I trust.

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Wed Jun 12, 2019 2:34 pm
by DaveMcW
eradicator's code is wrong.

Test setup: 9 beacons, 17 speed-module-3, 1 speed-module-2. Assembling machine 3 with max productivity making advanced circuits.

It produces exactly 1 recipe per second, so power and energy should be equal.

Re: Challenge: maximum net energy from a blue belt of coal

Posted: Wed Jun 12, 2019 2:41 pm
by eradicator
DaveMcW wrote:
Wed Jun 12, 2019 2:34 pm
It produces exactly 1 recipe per second, so power and energy should be equal.
Not if you use productivity. The function calculates output-cycles, not recipe-duration cycles.
Koub wrote:
Wed Jun 12, 2019 9:29 am
No offense, but there was no use even trying : if you read thoroughly my post, you'll see that I didn't even account for the consumption of the beacons.
None taken. But if you read through my post thoroughly you'll see that i also did a calculation test with free-energy beacons. I'm not a mathematician, but i think you and @DaveMcW are forgetting that productivity does make the recipe cheaper and faster.

Just think about how a hypothetical assembler with +100% productivity but no other bonuses would behave.