Do Prod+Speed Beacon setups save energy?

[BlakeMW]

The question of this topic is, can you save energy by using prod3/speed3 setups? You might think, What? You crazy? Isn't it an obvious exchange: You put absurd amounts of electricity in and get some free items out? Doesn't it obviously require more electricity? Except it's also really expensive in terms of energy to make items. In this analysis I'll be comparing with eff1 modules for the simple reason that using eff1 modules is actually in general a good and solid strategy, while using eff2 and eff3 modules is energetically a very unsound strategy because a watt generated by solar/accu is so much cheaper than a watt saved by eff2.

I'll be using an Assembler 3 doing the Electronic Circuits recipe. The assembler has 4 Prod3 Modules, and benefits from 8 Beacons, which all effect 8 assemblers, so it ends up with +400% speed from the beacons and pays a 1/8th share of the energy cost.

Assembler 3 doing Electric Circuits

• Energy Use (+880%): 210kW * 9.8 = 2058kW + (480kW * 8 / 8) = 2538kW
• Crafting Speed (+340%): 1.25 * 4.4 = 5.5
• Productivity Bonus: 40%
• Circuit Creation Rate: 5.5 / 0.5 * 1.4 = 15.4/s
• Iron Plates Consumption Rate: 5.5 / 0.5 * 1 = 11/s
• Copper Wire Consumption Rate: 5.5 / 0.5 * 3 = 33/s
• Iron Plates From Thin Air: 11/s * 0.4 = 4.4/s
• Copper Cables From Thin Air: 33/s * 0.4 = 13.2/s

So in this analysis the setup is considered to create 4.4 iron plates and 13.2 copper wires out of thin air every second, so the question is how much energy would it take to mine, smelt and refine those resources? For purposes of mining and smelting we will combine the copper and iron: 4.4 + 13.2 / 2 = 11 plates/s

The first energy number is no modules, the second number is with a full load of eff1 modules:

• Electric Mining Drill: 90kW * 1.75s * 11/s = 1732.5 kW | 346.5kW
• Electric Furnace: 180kW * 1.75s * 11/s = 3465kW | 1386kW
• Assembler 3 (cables): 210kW * 0.5s / 1.25 * (13.2 /3 )/s = 369.6kW | 73.92kW
• Total: 5572kW | 1806kW

So the total energy cost to produce those resources afresh is 5.6MW (no modules) or 1.8MW (eff1 modules), also just for completeness sake if you use steel furnace and are using steam power exclusively then by smelting directly with coal you can treat the smelting step as consuming half as much "electricity fuel value" - 1732.5kW, for a total of 3.9MW.

The final question is, how many assembler 3's does it take to make 15.4/s Electronic Circuits without prod/speed, and how much energy does it require with and without eff1?

• Assemblers required: 15.4/s * 0.5s / 1.25 = 6.16
• Energy Used: 210kW * 6.16 = 1293.6kW | 258.7kW

We can now do a grand total and ranking in terms of comparative energy cost:

• Eff1 Modules: 1806kW + 259kW = 2065kW
• Prod3+Speed3 beacons: 2538kW
• (Steel Furnace: 3840kW + 1293kW = 5133kW)
• No Modules: 5572kW + 1293.6kW = 6866kW

The conclusion is that Prod3+Speed3 Beacons can save a whopping amount of energy compared with no modules at all, this is because the energy savings by pulling resources out of thin air rather than producing them the honest way is huge.
However the cheap and effective eff1 modules result in even better energy savings, although the advantage eff1 modules have in energy savings is very small for the high yield recipes like Electronic Circuits, Processing Units, Plastic, Science Pack 3 and so on.

Closing Remarks

Is it fair to compare Prod3+Speed3 Beacons, which is really expensive, with eff1 modules, which is really cheap? Yes, it is because we are only comparing energy consumption, not total effect. With Prod3+Speed3 the reduced energy consumption is actually a fringe benefit, we don't build productivity modules to save energy, we do it to reduce the number of mines (etc) required and simplify logistics, that's why we pay the whopping costs for Prod3 modules, the energy savings are a freebie on top of that. So I'm comparing the side effect of something expensive, with the primary effect of something inexpensive.

Also how do the numbers work out for other recipes? Refer to the post Productivity Module Math. It's roughly proportional to to the ranking in the that list although there are some subtleties. For example Advanced Circuits are very energy intensive as they take a lot of assembly time to make. So Processing units, despite being a little lower on the list than Electronic Circuits, probably has a higher energy saving just because the intermediates required suck up more factory assembly time. It is plausible that in the case of recipes like Processing Units and Science Pack 3 you might use even less energy with Prod+Speed than with eff1, simply by saving so much assembly time on the intermediates.

[Qon]

It is plausible that in the case of recipes like Processing Units and Science Pack 3 you might use even less energy with Prod+Speed than with eff1, simply by saving so much assembly time on the intermediates.

This is the most interesting part. When do you save more energy with PM+SM than EM? In the silo it's obviously better with PM, but where do you draw the line if you want to minimize energy used?

Currently I'm thinking that I will use PM for anything made from plates/oil. Furnaces and miners don't have 4 slots so the PM+SM combo isn't as strong. And with increasing richness by distance making resource patches last longer isn't as important. The small effect of prod modules in miners and furnaces isn't worth it if you can go a bit further instead. So the goal for me is reducing the number of mining outposts and energy used. I'm fine with using some small bit of extra energy if it reduces outpost building time.

[Case]

You also should consider that beacons eat energy even when not used - so some sort of "power off" mechanic is needed to take use of this energy save.

Interesting experiment, though.

[BlakeMW]

You also should consider that beacons eat energy even when not used - so some sort of "power off" mechanic is needed to take use of this energy save.

Interesting experiment, though.

Whether or not you need to take into account idle beacons depends on whether you are using solar or steam. When using solar/accu you simply need to size your power generation to your factory's maximum consumption, and if the consumption falls below that maximum the excess electricity is simply lost, idle beacons literally cost you nothing and cause no additional pollution.
With steam you also need to size your generation to maximum consumption, but lower consumption can save a bit of fuel and reduce pollution a bit. What you need to consider is if the coal savings are worth the effort of switching off the beacons, which is largely going to depend on how much idle time you plan to have vs the map's coal richness.

[Case]

Ah, I see - fair enough.

[MeduSalem]

Wait... While reading through your calculation BlakeMW I had the thought of that you didn't compare the setups entirely correct.


Setup 1: PM3 + SM3 in beacon combination

Assembler 3 doing the Electric Circuits:

• Energy Use (+880%): 210kW * 9.8 = 2058kW + (480kW * 8 / 8) = 2538kW
• Crafting Speed (+340%): 1.25 * 4.4 = 5.5
• Productivity Bonus: 40%
• Circuit Creation Rate: 5.5 / 0.5 * 1.4 = 15.4/s
• Iron Plates Consumption Rate: 5.5 / 0.5 * 1 = 11/s
• Copper Cable Consumption Rate: 5.5 / 0.5 * 3 = 33/s
• Iron Plates From Thin Air: 15.4/s - 11/s = 4.4/s
• Copper Cables From Thin Air: 15.4/s * 3 - 33/s = 13.2/s

I agree to that. But now:

Assembler 3 doing the Copper Cables:

• Assemblers required: 33/s * 0.5s / 1.25 = 13.2
• Energy used: 210kW * 13.2 = 2772kW | 554.4kW

Electric Furnace doing the Iron/Copper Plates:

• Furnaces required: (11/s + 33/s * 0.5) * 1.75 = 48.125
• Energy used: 180kW * 48.125 = 8662.5kW | 1732.5kW

Electric Mining Drill doing the Iron/Copper Ore:

• Drills required: (11/s + 33/s * 0.5) * 1.75 = 48.125
• Energy used: 90kW * 48.125 = 4331.25kW | 866.25kW

Total:

• Energy Used if no Modules before Electronic Circuits: 2538kW + 2772kW + 8662.5kW + 4331.25kW = 18303.75kW
• Energy Used if Efficiency Modules 1 before Electronic Circuits: 2538kW + 554.4kW + 1732.5kW + 866.25kW = 5691.15kW

Setup 2: No Modules | Setup 3: Efficiency Modules 1

We want to match the 15.4/s Electronic Circuit output of the PM3+SM3... so:

Assembler 3 doing the Electric Circuits:

• Assemblers required: 15.4/s * 0.5s / 1.25 = 6.16
• Energy Used: 210kW * 6.16 = 1293.6kW | 258.7kW
• Crafting Speed (+0%): 1.25 * 1 = 1.25
• Productivity Bonus: 0%
• Circuit Creation Rate: 15.4/s
• Iron Plates Consumption Rate: 15.4/s
• Copper Cable Consumption Rate: 3 * 15.4/s = 46.2/s
• Iron Plates From Thin Air: None
• Copper Cables From Thin Air: None

Assembler 3 doing the Copper Cables:

• Assemblers required: 46.2/s * 0.5s / 1.25 = 18.48
• Energy used: 210kW * 18.48 = 3880.8kW | 776.16kW

Electric Furnace doing the Iron/Copper Plates:

• Furnaces required: (15.4/s + 46.2/s * 0.5) * 1.75 = 67.375
• Energy used: 180kW * 67.375 = 12127.5kW | 2425.5kW

Electric Mining Drill doing the Iron/Copper Ore:

• Drills required: (15.4/s + 46.2/s * 0.5) * 1.75 = 67.375
• Energy used: 90kW * 67.375 = 6063.75kW | 1212.75kW

Total:

• Energy Used if no Modules at all: 1293.6kW + 3880.8kW + 12127.5kW + 6063.75kW = 23365.65kW
• Energy Used if Efficiency Modules everywhere: 258.7kW + 776.16kW + 2425.5kW + 1212.75kW = 4673.11kW

Ranking list is as following

1. Efficiency Modules everywhere: 4673.11kW
2. PM3/SM3 AND Efficiency Modules 1 before Electronic Circuits: 5691.15kW
3. PM3/SM3 AND no Modules before Electronic Circuits: 18303.75kW
4. No Modules at all: 23365.65kW

With other words the EM1 only version is still best... and the one PM3/SM3 combination can only compete because is also using EM1s before the Electronic Circuits stage.

That said if I didn't make any mistakes in the reverse calculation.

Honestly it would be interesting to know how much more energy could be saved if using PM3/SM3 stuff before the Electronic Circuits. Probably not enough to outperform the EM1s.

But I can imagine that for recipes with multiple layers of PM3/SM3 combinations, like for example 1-minute rocket launch mega bases might have enough energy savings through the process to outperform a setup using only EMs.


Maybe I will come back to it and include PM3/SM3 also to Copper Cable, Smelting and maybe Mining Drills. Should actually be a simple matter of dividing by 1.4 and 1.2 respectively and adjusting the power consumption to the use of PM3s+SM3s in beacons accordingly.

[BlakeMW]

You calculated the total energy, while I calculated the energy requirements to make the free items.

In principle the difference in energy between each approach should be identical for both calculations - although you have given the furnaces eff2 modules in all, but I think you intended to only give it eff1 for the first calculation.

[MeduSalem]

You calculated the total energy, while I calculated the energy requirements to make the free items.

Yeah, your calculation basically shows how much energy the "free items" would have used if they were done by regular means.

But that doesn't really demonstratively compare the energy consumption of varying setups all capped at a throughput of 15.4 EC/s... which is a much more illustrative comparison because it directly shows numbers that can be read inside factorio in your power production statistics.

Basically after my calculation one could go ahead and plan the power plant accordingly and there shouldn't be any surprises.

In principle the difference in energy between each approach should be identical for both calculations - although you have given the furnaces eff2 modules in all, but I think you intended to only give it eff1 for the first calculation.

True, but not scaled realistically.

In your ranking the No Module setup is 3.3 times as bad as the EM1 variant.
In my setup the No Module setup is 5 times as bad as the "best" EM variant. (I know best is relative, others may find EM2s not worth the trouble)

But the striking discrepancy is in the PM3/SM3 combination that shows off as if it would be almost as good as EMs but in reality not even close because to make the PM3/SM3 as good would either mean using EM2s everywhere before ECs or maybe using PM3/SM3 everywhere before (which would be interesting to research).

That said you are right, I used EM2s in the calculation because that's how I would roll if using EMs at all. Why not work at the 20% cap if it's possible... In the long run the investment would pay off (yeah I am playing maps at least a couple hundred of hours).

[BlakeMW]

But the striking discrepancy is in the PM3/SM3 combination that shows off as if it would be almost as good as EMs but in reality not even close because to make the PM3/SM3 as good would either mean using EM2s everywhere before ECs or maybe using PM3/SM3 everywhere before (which would be interesting to research).

The thing is in my analysis I wanted to look at a single assembler in isolation and in particular devise a way of putting an energy value on the the free items created by a prod module, so you can compare "the prod module adds X energy to the consumption of the assembler, and it saves Y amount of energy by creating free items".

I'm not saying that's a better approach than a full chain or total energy cost analysis, in fact I think both approaches should be taken, and I am happy you decided to do a total energy cost analysis.

That said you are right, I used EM2s in the calculation because that's how I would roll if using EMs at all. Why not work at the 20% cap if it's possible... In the long run the investment would pay off (yeah I am playing maps at least a couple hundred of hours).

In the long run the investment would only pay off if you are using steam power. Otherwise since there is no ongoing cost to solar/accu it never pays off.

• The capital cost of eff2 modules (per watt) is about 5x higher than the capital cost of solar/accu (per watt)
• A watt generated by solar/accu costs nothing.
• A watt saved by an eff2 module costs nothing.

Now tell me, how exactly can the eff2 module ever pay off its higher capital cost? If every watt generated by solar/accu cost something, and every watt saved by eff2 cost nothing, then it could pay off eventually. As it is, eff2 and eff3 modules can never pay off vs solar accu, only vs steam, so it can only pay off under conditions of self-imposed restrictions where the absolutely superior option is restricted.

If you really care about pollution, then the payoff time in terms of pollution alone isn't too bad (that is, the pollution to create an eff2 module, vs the pollution the eff2 module saves in a furnace), I think I calculated it to be a few hours, but it's probably more like 10 hours when compared with an eff1 module. But when you consider the resource cost of getting that pollution reduction, vs the benefit you get from that pollution reduction then the payoff time would be much longer (it actually could be quantified in terms of biters created by spawners absorbing pollution, vs the resource cost to kill those biters as in bullets or more reasonably investment in solar/accu/laser... maybe another time...)

[MeduSalem]

In the long run the investment would only pay off only if you are using steam power. Otherwise since there is no ongoing cost to solar/accu it never pays off.

• The capital cost of eff2 modules (per watt) is about 5x higher than the capital cost of solar/accu (per watt)
• A watt generated by solar/accu costs nothing.
• A watt saved by an eff2 module costs nothing.

Now tell me, how exactly can the eff2 module ever pay off its higher capital cost? If every watt generated by solar/accu cost something, and every watt saved by eff2 cost nothing, then it could pay off eventually. As it is, eff2 and eff3 modules can never pay off vs solar accu, only vs steam.

Funny thing... I am a Steam only guy. So my perspective was from the point of view using Steam engines of course.

I never use Solar power because I think that it doesn't add any worthwhile gameplay value. It's basically plop&forget... Free energy forever without any maintainence or other things to worry about (That in a game that is about building and maintaining a factory... the irony is astonishing on that part). Even if the accumulators run out during the night you are guranteed that with sunrise the factory will start working again so you aren't even punished for a complete power failure. But I think I had the discussion of why I don't really like Solar Power at all already enough over in the "Solar Power less of a no brainer" thread in the balancing section. So no need to go all over that again.

So apart from that you are right... EM2s with Solar Power don't make much sense with your reasoning and I see why you don't like using them. You can't save on something that already is available infinitely and where the only thing balancing the two things are initial crafting costs.

That said using EM2s in combination with Solar Power would reduce the Solar Power plant a little bit if one really cares about space being wasted rather than resources being wasted... But then I doubt that someone like that will use Solar Power in the first place due to how Steam Power is much more compact for a compareable power output.

[BlakeMW]

I'm also a mostly steam guy, often use solar/steam. But the game can't really be balanced around self-imposed rules. Also I mainly use steam for comparative advantage, solar and especially accu is frigging expensive upfront, accumulators cost *so much oil* which you can use for other things, like making early prod/speed setups. In fact I think many player's love of accus is what causes common perception that oil is scarce.
But the same logic which causes me to dislike making accumulators - that they are so expensive for what they do - applies quadruply to high lvl eff modules.

[Qon]

That said using EM2s in combination with Solar Power would reduce the Solar Power plant a little bit if one really cares about space being wasted rather than resources being wasted... But then I doubt that someone like that will use Solar Power in the first place due to how Steam Power is much more compact for a compareable power output.

As I said in the beacon energy cost thread: Why are you using EM2? Assembly machine 3 can take 3 EM1 for the same power reduction as 2 EM2. Only electric furnaces (of the machines worth considering) have a limit of 2 module slots. If you aren't combining your EM with SM then EM 1 is fine.

[MeduSalem]

As I said in the beacon energy cost thread: Why are you using EM2? Assembly machine 3 can take 3 EM1 for the same power reduction as 2 EM2. Only electric furnaces (of the machines worth considering) have a limit of 2 module slots. If you aren't combining your EM with SM then EM 1 is fine.

Yeah well, you can basically imagine that 3 EM1s are inside an AM3 if you like to save on the resources.

But you are right... It might even make sense to use EM3s when you are using SM3s inside AM3s too... but I never really experimented much with such exotic setups, so I don't know.

[Qon]

As I said in the beacon energy cost thread: Why are you using EM2? Assembly machine 3 can take 3 EM1 for the same power reduction as 2 EM2. Only electric furnaces (of the machines worth considering) have a limit of 2 module slots. If you aren't combining your EM with SM then EM 1 is fine.

Yeah well, you can basically imagine that 3 EM1s are inside an AM3 if you like to save on the resources.

But you are right... It might even make sense to use EM3s when you are using SM3s inside AM3s too... but I never really experimented much with such exotic setups, so I don't know.

Not just save resources. It's cheaper AND faster.

Simple math for simple EM + SM mixing:
EM3 saves 50% each. 3xEM3 saves 150%. The limit is -80% so we have a bonus 70 more than we need.
SM3 increases speed by 50% and electric usage by 70%
So 3 EM3 + 1 SM3 reduces electric usage by 80% (the limit) and increases speed by 50% (which makes it more compact and energy efficient for the same production values).
Expensive, but long term it's better. Remember that higher crafting speed is saving you energy. And that's also why AM3 is faster and more energy efficient than AM3.

[hi_impact]

Thanks everyone for this engaging conversation.

I do have a pleb question though - assuming I have the basic 3-2 Electronic Circuit setup, space-wise horizontally it looks like this:
Copperplate input belt -> 3x vertical copper wire -> 2x vertical electronic circuit -> circuit output belt -> ironplate input -> mirrored on other side

How exactly do you get 8 beacons around that? Was the 8 beacon example a thought experiment, or some bot build? Maybe I just can't wrap my head around it but 8 beacon EC setup that is realistic using belts.

[MeduSalem]

How exactly do you get 8 beacons around that? Was the 8 beacon example a thought experiment, or some bot build? Maybe I just can't wrap my head around it but 8 beacon EC setup that is realistic using belts.

Well if you really go for the beacon setup I guess bots are the only real viable choice then because of the throughput bottleneck introduced by the belts. That's why I am using bots.

But that said it's possible with belts though, but... not very aesthetically pleasing in my opinion. I hate belt braiding, but I'm especially unfond of the buffer chests for the copper wire, but probably better than outputting to the belt:

0.13 - Electronic Circuits + Beacons + Belt.jpg (133.43 KiB) Viewed 10271 times

Above setup would be good for using PM3s also in Copper Wire assemblers. The 140% productivity almost makes up for the ratio of 1.5... It's a trade off I'm willing to take because the alternative would mean having a belt or bots bring in the wire... So not going to do that.


That said... even the stack inserters have problems keeping up with the production speed, especially when grabbing from belt. Completely ridiculous.

[BlakeMW]

With copper wire I usually prefer to cheat and use robots - just moving from a passive provider to a requester chest next to it or perhaps the next row over. Robots truly excel at moving stuff extremely short distances. I'm not a big fan of bot bases, but I do like to use them where they truly excel.

In my latest setup I used this for my green circuits supply to red/blue/modules, nearly 2 express belts in, nearly 1 express belt out:

(btw it's not a perfect setup,it started as prod1+speed1 beacons and evolved from there - only the middle EC assembler actually operates at full speed)

The main challenge is having enough output inserters. With prod3 modules the EC recipe consumes 4 items and outputs 1.4 items so ideally you need in:out inserters in a 2.85:1 ratio: with only 6 inserter slots that means using 3:2 or 4:2. In practise with 2 stack inserters on output and 3 on input the inserters and assemblers are almost perfectly matched at +340% speed - if it were any faster the inserters just couldn't keep up, but happily they can just barely keep up at that maximum practical speed. Stack Inserters are a bit faster to/from chests but they aren't too bad to/from express belt, it's still fast enough to work (but it doesn't work well enough for copper wires, which is why I have the copper wires using logistic chests).

edit:

A more optimized setup

This setup outputs the green circuits to both sides of the belt. Basically consumes nearly an express belt of feedstock and puts out nearly an express belt of green circuits.

[MeduSalem]

With copper wire I usually prefer to cheat and use robots - just moving from a passive provider to a requester chest next to it or perhaps the next row over. Robots truly excel at moving stuff extremely short distances. I'm not a big fan of bot bases, but I do like to use them where they truly excel.

There would be an additional benefit of using bots to transport the Copper Wire I didn't give much thought about until now... Since the conversion ratio of 1.4 Copper wire produced by an AM3 with PM3 doesn't match the 1.5 required copper wire of the EC AM3 one could strategically place one additional AM3 assembler producing copper wire somewhere. It would produce the missing 0.1 Copper wire, meaning one would have one such extra copper wire AM3 every other 14 EC AM3s... and the ratio would be count perfect again, meaning the EC AM3s wouldn't stall every once in a while to wait for the missing copper wire derived from the conversion ratio discrepancy.

edit:

[...]

This setup outputs the blue circuits to both sides of the belt. Basically consumes nearly an express belt of feedstock and puts out nearly an express belt of green circuits.

You mean green circuits right?

But apart from that your second setup is really better... It basically balances the circuit output to both sides of the belt "evenly" (depending on which of both output inserters becomes the dominant one).

[BlakeMW]

There would be an additional benefit of using bots to transport the Copper Wire I didn't give much thought about until now... Since the conversion ratio of 1.4 Copper wire produced by an AM3 with PM3 doesn't match the 1.5 required copper wire of the EC AM3 one could strategically place one additional AM3 assembler producing copper wire somewhere. It would produce the missing 0.1 Copper wire, meaning one would have one such extra copper wire AM3 every other 14 EC AM3s... and the ratio would be count perfect again

Yep, it's one reason I like to bot copper cable. It's also nice to be able to request cable from the network to mass produce beacons.

You mean green circuits right?

But apart from that your second setup is really better... It basically balances the circuit output to both sides of the belt "evenly" (depending on which of both output inserters becomes the dominant one).

derp. And yes, the second setup is an intelligently designed one. The first one... just organically evolved.

[hi_impact]

Thanks Medusalem and Blake. Appreciate those builds. Crazy how much that one EC factory outputs. Never played too much with beacons. Fun stuff.