Does a beacon consume too much energy?

[siggboy]

Albeit it will sound almost like heresy to all the 3:2 perfect ratio people... I am using 4 Productivity Modules in Both the EC machine and the Cable machine behind it... saves a lot of resources.

The PMs in the EC machine save a lot of resources, the cable machine is already less important (but still good).

1:1 setups are a lot more compact than 3:2 and they're easier to integrate into blueprints where the EC machine directly feeds into something else (like AC or PU assemblers).

People who stick to 3:2 without thinking about the alternatives are misguided.

(Also I've made two other points editing that post above while you were replying, sorry about that.)

[MeduSalem]

Albeit it will sound almost like heresy to all the 3:2 perfect ratio people... I am using 4 Productivity Modules in Both the EC machine and the Cable machine behind it... saves a lot of resources.

The PMs in the EC machine save a lot of resources, the cable machine is already less important (but still good).

1:1 setups are a lot more compact than 3:2 and they're easier to integrate into blueprints where the EC machine directly feeds into something else (like AC or PU assemblers).

People who stick to 3:2 without thinking about the alternatives are misguided.

I think both EC/cable machine are quite noteworthy...

If I understand correctly one requires only 1/1.4 = 71% the resources for EC when using PM3s... further using productivity in cable too reduces the copper consumption to (1/1.4)^2 = 51% of before. So another gain. With Furnaces doing their PM job too... it decreases to (1/1.4)^2*(1/1.2) = 42%.

So the effect stacks quite noticably... which in return means I have to do less mining and less tedious outpost building.

(Also I've made two other points editing that post above while you were replying, sorry about that.)

Yeah I noticed...

[...]so I don't understand why you talk about it so much

Well I am only that crazy about the EM stuff when I hear "energy saving" in combination with Beacons... That's all.





Finally back to topic... I guess.

I think the energy consumption of beacons is "okay"... for the most part.

One thing that currently lacks is the ability to shut down individual beacons via circuit network if they are not needed.

The power switches can only do so much... mostly shut down an entire array of machines, which means that if you shut down arrays of beacons you have unpredictable energy spikes every now and then, which may be undesired. Would wish for a way to gradually shut them down one by one than just "ba dum tss" effects all the time.

I also experimented with swapping modules in/out of the beacons but... man that's tedious and requires additional building space which isn't always available in such tight spots.

[siggboy]

One thing that currently lacks is the ability to shut down individual beacons via circuit network if they are not needed.

The power switches can only do so much... mostly shut down an entire array of machines.

If you build modular mini-factories, which is arguably the best design principle for mega-bases, then shutting down the beacons along with the machines that they boost is entirely sufficient. Why would you want to shut down single beacons and leave the machines running?

Either the machines are doing work (then you want the beacons), or they don't, and then you can shut them down entirely, along with the beacons.

I also experimented with swapping modules in/out of the beacons but... man that's tedious and requires additional building space which isn't always available in such tight spots.

I actually liked your module swapping hack, but I agree that it was probably not really practical.

[MeduSalem]

If you build modular mini-factories, which is arguably the best design principle for mega-bases, then shutting down the beacons along with the machines that they boost is entirely sufficient. Why would you want to shut down single beacons and leave the machines running?

Either the machines are doing work (then you want the beacons), or they don't, and then you can shut them down entirely, along with the beacons.

Well that's why I edited the post once to clarify why...

But I'll elaborate a bit more:

While experimenting around with using Rocket Fuel in Boilers (viewtopic.php?p=180397#p180397 & viewtopic.php?p=180442#p180442) I found that using Power switches is really helpful to minimize Beacon leakage in the production steps.

But I also realized that once the Beacons+Assemblers etc come back online as a batch it creates huge sparks in the power consumption... which required an awful lot of Accumulators to catch (that or a hot water storage, not sure about that yet).

Which is why a more "gradual" approach would be more desireable... at least for me. May be a personal taste of mine of course but I hate unpredictible sparks.

I also experimented with swapping modules in/out of the beacons but... man that's tedious and requires additional building space which isn't always available in such tight spots.

I actually liked your module swapping hack, but I agree that it was probably not really practical.

Yeah, it took a lot of space... and I did it for the same reason as stated above... to match the resource consumption more gradually than "everything is on" or "everything is off".

[BlakeMW]

On the topic of whether Prod3/Speed3 can save energy.

For me eff2 doesn't make sense at all, eff2 is so damn expensive per kW saved that it just doesn't make sense to use them when you can make Solar/Accu. But on the other hand eff1 is a little bundle of overpowered goodness and in many machines takes the power usage down to the -80% limit. So I'm going to compare with eff1 since eff1 makes sense.

Take Electronic Circuits for the purpose of comparison, using 4x Prod3 stretches Copper and Iron by 40%, and in an infinite alternating rows setup it consumes 2.6MW (including beacons).

In this setup the Assembler 3 has a crafting speed of 5.5 and a productivity boost of 1.4, every second it consumes 5.5 iron plates and 8.25 copper plates, and in effect you get 2.2 iron plates and 3.3 copper plates for free - below I'll just call it 5.5 plates when it doesn't matter what kind of plate it is.

We need to consider the energy to mine and smelt 5.5 plates/s, I'm going to call it 1.75s for the mining:

• Mining Energy Cost: 90kW * 1.75 * 5.5 = 866kW
• Smelting Energy cost: 180kW * (3.5 / 2) * 5.5 = 1732kW
• Copper Wire energy cost: 210kW * (0.5 / 1.25) * 3.3 = 277kW

So in total the machine is saving us 2.9MW of mining/smelting/crafting energy per second. But if we put eff1 modules into those machines the number is reduced to 866kW * 0.2 + 1732kW * 0.4 + 277kW * 0.2 = 921kW.
(note that for the baseline factory, if it is purely steam powered and using steel furnaces you can halve the smelting energy cost for purposes of comparison since the steel furnace is twice as efficient when consuming fuel directly)

So the maximized assembler is saving us 2.9MW in a factory with no modules, or 921kW in a factory with universal eff1 modules.

Next step: The Prod3/Speed3 assembler consumes 2.58MW and the question is how much energy would it require to match that output using eff1 assemblers? We need to match an effective crafting speed of 5.5 * 1.4 = 7.7

So we need 7.7 / 1.25 = 6.16 Assembler 3, without modules these would consume 1294kW but with eff1 this is reduced to 259kW.

Energy to produce an equivalent circuit output from a pure eff1 setup:
259kW + 55kW + 866kW = 1.1MW


So the final ranking is:

• No Modules: 4.2MW
• Prod3/Speed3: 2.6MW
• Eff1: 1.1MW

In conclusion the energy savings from getting materials out of thin air are nowhere near as large as the savings from using eff1 modules to mine/smelt/process extra resources. Nevertheless compared with the baseline, the Prod3+Speed3 beacon setup actual does save energy for high yield recipes, so it's completely accurate to say that it saves energy - although it doesn't save as much energy as an eff1 setup.

And there is one recipe which is so high yield that prod3 modules always save energy - Rocket Part. The rocket silo consumes resources at a rate 12x greater than electronic circuits in Assembler3, and the rocket silo only consumes 440kW when crafting. Considering 4x Prod3 modules make the whole rest of the supply chain of your factory 40% bigger at the cost of only a few hundred kW there is no competing with them.

[MeduSalem]

So the final ranking is:

• No Modules: 4.2MW
• Prod3/Speed3: 2.6MW
• Eff1: 1.1MW

In conclusion the energy savings from getting materials out of thin air are nowhere near as large as the savings from using eff1 modules to mine/smelt/process extra resources. Nevertheless compared with the baseline, the Prod3+Speed3 beacon setup actual does save energy for high yield recipes, so it's completely accurate to say that it saves energy - although it doesn't save as much energy as an eff1 setup.

And there is one recipe which is so high yield that prod3 modules always save energy - Rocket Part. The rocket silo consumes resources at a rate 12x greater than electronic circuits in Assembler3, and the rocket silo only consumes 440kW when crafting. Considering 4x Prod3 modules make the whole rest of the supply chain of your factory 40% bigger at the cost of only a few hundred kW there is no competing with them.

So basically the more layers of PM3s can be used the more savings on required intermediate items/resources and thereby also less and less energy spend on producing them...

Too bad that most supply chains are too short to make a profit of that stacking behaviour. Otherwise there might be more breakeven points where a PM3+SM3 combination is better than using EMs when it comes to energy usage.

That said... one still saves a lot of resources which would have to be mined (with tedious outpost building wasting your time etc), while energy can be produced pretty much infinitely.

[siggboy]

Lovely analysis, Blake.

I already said above that if energy saving is your main concern then you put Eff into anything and that's it.

However, energy saving is not everything, you also want to make compact factories, because that is usually easier to build and maintain (yes, I know we have "infinite" space, but practicality plays a role, too).

Likewise, the resource savings make the factory easier to run on a large scale, because you need fewer outposts, fewer trains, fewer everything for the same amount of rockets per minute.

(Let's be honest here, ultimately we're optimizing for a good ratio of fun-to-rockets-per-minute, and nothing else ).

[Qon]

@MeduSalem that was a long post with a lot of wrong. I'll try to avoid reapeating too much of what siggboy says.

Beacons reduce overall energy used by a lot if used properly.

Compared to which setups?

There are only three possible setups to compare:

• No Modules in machines.
• EM2s in all machines.
• PM3s in all machines + SM3s in Beacons.

Anything else doesn't really make much sense.

You question my statement with an example of why I'm right. That one was easy to answer.

And I've given another example in the steam thread. If your goal is a certain number of items/s then productivity in the end of your production chain will reduce your whole factory by 30% and so also your energy needs decrease by almost 30%. Naturally you use speed beacons there too. Rocket silo is one good example where this effect is very real. Even using no beacons at all would be better than not using productivity modules because of the big reduction in inputs required.

3xEM1 + AM3 is cheaper, faster and uses less much less energy per craft than than 2xEM2 + AM2. EM2's are fine for electric furnaces but shouldn't really be used anywhere else for most cases.
3xEM3 + 1xSM3 in a AM3 uses even less energy per craft and you get even faster prodoction!

If you meant PM3+SM3 in comparison to using PM3s only then you are right and no real need to read further. Energywise PM3 only is worse than Beaconized (an efficient setup using beacons is almost twice as efficient than using PM3s without beacons from my measurements)... which is why I don't even count PM3 without beacons as a "legal" option to play with anymore.

Well I did say "if used properly".

But you can use productivity without speed beacons. If your input isn't coming in fast enough and you have energy to spare then it's one such case. Completely legal.

Another example would be a bot factory. More compressed factories requires less bot travel distance and allows more space for extra roboports so robots don't have to waste as much energy on detours for charging. And faster output increases probability of output chests being filled with more than 1 item before a robot comes to pick it up, thus increasing the emount you take advantage of the carrying capacity and decreasing robot use. And smaller factories requires less roboports for area coverage. A factory with EM instead of productivity will be so big that robot travel distance could become one of the more expensive energy sinks. If you save energy overall depends on a lot of factors and how large your factory is and the chaotic nature of robots. Anyting less than a megafactory is too small for it to be a net gain, and a megafactory will be too hard to compute. But maybe 4xEM3 in AM3 + 2xSM3 beacons could actually be more energy efficient than just using EM in AM3 without beacons if your bot factory is large enough. Of course using belts could be even better for your energy needs, but then we aren't just adding speed beacons any more.

Back to the topic... Sure with EM2s you need an awful lot more machines to get close to the throughput of a well beaconized setup, but the EM2s will still reduce the energy consumption to such a low level that the beaconized setup can't compete with the energy consumption.

How much energy does a factory use with efficiency modules only to launch 1 rocket? No beacons or robots counted. Shouldn't be too hard to calculate compared to a productivity factory. Might try later. Or if you know of a factory that is only using EM and does so everywhere, have a screenshot of electric usage when the factory is only used for launching rockets, know time/launch and that the factory isn't producing from buffers then simple multiplication would give us a fairly accurate number. I'm curious.

In my experience Beaconized Setups only reduce the footprint by a lot (apart from the free item every now and then). But on the other hand the Beacons and Module effects increase the energy consumption so much that you trade space gain for energy consumption. Basically you only increase the Production/Area ratio on account of additional energy consumption.

And decreased material use, which are otherwise bought with energy, space and time.

Many others have posted since I started typing this reply. Whatever.

[Qon]

If you build modular mini-factories, which is arguably the best design principle for mega-bases, then shutting down the beacons along with the machines that they boost is entirely sufficient. Why would you want to shut down single beacons and leave the machines running?

For megabases with beacons you want to have one big block of beacons and assembly machines producing everything. Serparation increases energy wasted by beacons that can't cover 8 machines. And then you can't shut down beacons for individual machines any more without also affecting the production of other machines producing something else while sharing beacons. It's also superior for making expansions.

On the other hand if you do it at the correct ratio you either have all machines running or none, so you can turn off the whole megabase with an energy switch instead.

[siggboy]

If you build modular mini-factories

For megabases with beacons you want to have one big block of beacons and assembly machines producing everything.

With "modular mini-factories" I mean one of these things:

AC Module

ac-module.PNG (881.41 KiB) Viewed 3520 times

That design is by DaveMcW, he actually uses interleaved belts to get the output from the AC assemblers and puts it right into another one of these beauties making PU, that would be right below and also in range of the beacons power ing the AC assemblers. It's really very nice, and tileable. I refuse to do interleaved belts so I'm using robots here for the output. But that screenshot is on a sandbox map, not in a real factory.

It's also slightly broken in this screenshot because the inserters can't pick up everything from the blue belts (I was experimenting with this, had no perfect version at hand for taking a screenie right now), but I guess the principle is clear.

This is far better than putting all the EC and Plastic on belts and it still gives you very good beacon coverage, except at the edges, but any setup has this limitation.

[Qon]

It's also slightly broken in this screenshot because the inserters can't pick up everything from the blue belts (I was experimenting with this, had no perfect version at hand for taking a screenie right now), but I guess the principle is clear.

This is far better than putting all the EC and Plastic on belts and it still gives you very good beacon coverage, except at the edges, but any setup has this limitation.

If your inserters can't pick the items up you aren't producing enough.

You ARE putting all EC and plastic on belts.

[siggboy]

If your inserters can't pick the items up you aren't producing enough.

Nah, that's not the problem in that screenshot. Inserters can't pick up from underneath the "hood" of an underground express belt, unless it's from an incoming belt (not from the flanks/sides). Not even fast inserters. So these belts need to be red belts (and they in fact are, in DaveMcWs original design). I was messing with the setup and replaced them with blue belts, although I actually knew it wouldn't be working for that reason.

You ARE putting all EC and plastic on belts.

Yeah, don't be silly, I was talking about "bus belts" that you run through the factory, to get all the EC from one location to another where they are consumed. The EC in this design are produced locally, the belts are only to get from from one machine into the next.

[Qon]

If your inserters can't pick the items up you aren't producing enough.

Nah, that's not the problem in that screenshot. Inserters can't pick up from underneath the "hood" of an underground express belt, unless it's from an incoming belt (not from the flanks/sides). Not even fast inserters. So these belts need to be red belts (and they in fact are, in DaveMcWs original design). I was messing with the setup and replaced them with blue belts, although I actually knew it wouldn't be working for that reason.

Yes it is. If the items stop then you can even use burner inserters to pull off from blue underground belts from the sides. And the items stop when they hit the end of the belt or another item on the belt. If you have enough production then items will build up on the belt enough for all inserters to pick up items as long as you aren't looping the belts back. If the items don't build up enough for items to stop then the last consumers on the belt are eating the materials fast enough to prevent buildup.

[siggboy]

These modules produce enough of everything to work at close to 100% saturation. Making more just to back up the belts (if that would even work in that setup) is not the solution. The solution is to use red belts.

[Qon]

These modules produce enough of everything to work at close to 100% saturation. Making more just to back up the belts (if that would even work in that setup) is not the solution. The solution is to use red belts.

You don't overproduce to back up the belt. If the supply is enough for all consumers then the last consumers can't consume fast enough to prevent the belt from backing up. Are you saying that the items reach the last assemblers and everything works at full speed with red underground belts? Seems implausible, with the same production the materials have to go somewhere. The products can't just disappear. Maybe the inserters miss some items so that they reach the last machines with red undergrounds, but then the first assemblers wouldn't be working 100% of the time any more.

[siggboy]

If the belt content is moving then the inserters can't pick up the items from underneath the belt cover (from the sides). So it will at least slow down production, even if the belt is saturated, because it becomes "un-saturated" as soon as items are picked up and the resupply moves in from the back.

The inserters need to be able to reliably pick up a moving item, otherwise you will reduce the throughput of the entire module, as machines will idle when they shouldn't.

You can remove all of these delays by simply using red belt, because that's still more than fast enough to bring in the supply, without the downside of slowing down the insertion process (as I've said before, the original design used red belt, they're fast enough, it was my mistake to replace them with blue).

Also, if your ratios are correct, then everything that is put on the belt upstream will be fully consumed downstream, and the belt will never, ever be completely backed up, while the machines are running.

[BlakeMW]

I actually realized my earlier calculations were off by a factor of 2 because I had the electronic circuits at 1/s instead of 0.5/s, in fact prod3+speed3 beacons is even more energy-friendly than I had calculated (but still not quite as good as eff1). But in any case I made a new discussion with the correct calculations: viewtopic.php?f=5&t=28592

[Qon]

If the belt content is moving then the inserters can't pick up the items from underneath the belt cover (from the sides). So it will at least slow down production, even if the belt is saturated, because it becomes "un-saturated" as soon as items are picked up and the resupply moves in from the back.

The inserters need to be able to reliably pick up a moving item, otherwise you will reduce the throughput of the entire module, as machines will idle when they shouldn't.

That's like saying that the last machines will not run at full speed with regular belts/red UG because when the first inserters grab the items from the belt there will be nothing left for the last red circuit producer where the belt is desaturated, the the last inserters will fail to pick up items. Which is obviously wrong. After a while the input buffers of the first assemblers will fill up and cables will flow past and reach the last assemblers. It needs a warmup time where the first ones are gets priority in the beginning. With UG belts it's the opposite, the last ones gets priority but after buffers fill up everything gets distributed equally.

Also, if your ratios are correct, then everything that is put on the belt upstream will be fully consumed downstream, and the belt will never, ever be completely backed up, while the machines are running.

Wrong. Lets say you have a copper cable machine producing for 8 advanced circuit assemblers (no PM, all at same speed, so ratio correct.) For the belt to not back up to the 2nd last assembler the last assemblers will have to be able to eat the entire cable output by itself. For the belt to not back up to the first assembler the 7 last assemblers have to be able to consume the entire output of the copper cable assembler.

Otherwise you have to explain where the extra cables are going.

I'm right, you are wrong. Sorry. And since you can't be bothered to test what I'm saying even though you don't know what you assert, I made a test setup for you with a screenshot to prove it. Enjoy. And yes I did empty the cable belt in the beginning, it still backed up. And the cable machine is running at 100% even though the belt has backed up.
If you want to actually run your own test then please let the system find equilibrium before you discard it. Just like you don't watch the production screen before the last assembler gets it's input when using red UG. If this is worse than red belts then it's less than 0.01% difference and using regular or red belts might be actually be worse by that margin.

I'm using stack inserters here because of speed modules. Thought that running it on highest speed possible would push my assertion to the limit. The red UG belt would be too slow in this exact setup though so you would have to slow the assemblers down a bit to make a direct comparison where the only difference is the color of the UG belts.
Do you want me to make a similar setup where I use burner inserters and show that they can pick up from a blue belt too or can we just assume that when I say something it's because I know I'm right and that I'm right when I think I am? q:

[siggboy]

I will look into it and run some tests myself. Right now what I take for fact is that inserters can not pick up at full speed from underneath a blue underground belt unless it's moving towards the inserter. And if full speed is required in order to keep the machine running at 100% then I don't see how to get around this.

If full speed is not required then it's a different story (and I'm not doubting your analysis, after all you've tested it in game).

Also in 0.13, because the inserters actually pick up 3 items from the belt (at full tech) instead of only one, that changes things a lot. So this discussion might be a bit moot in 0.13.

[Qon]

I will look into it and run some tests myself. Right now what I take for fact is that inserters can not pick up at full speed from underneath a blue underground belt unless it's moving towards the inserter. And if full speed is required in order to keep the machine running at 100% then I don't see how to get around this.

Also in 0.13, because the inserters actually pick up 3 items from the belt (at full tech) instead of only one, that changes things a lot. So this discussion might be a bit moot in 0.13.

I think they can pick up at full speed as long as the belt is compressed, because then it will fail to pick up one item but grab the next one instead and succeed because it has already stretched out far enough to be able to grab instanty. But it doesn't matter, because the items aren't moving on the belt in my picture when they are grabbing the items, because it backs up. If they can't grab the items then items picked up is less than items put down on the belt and then it backs up. Once it backs up they are sitting still and can easily be picked up. The assemblers have input buffers large enough and the cycles between items standing still and moving are short enough that no machine runs out of material.

Don't think 0.12 or 0.13 matters really. As long as the inserters can insert fast enough that should be enough. I said I used stack inserters so you know which version I used to test this. But burner inserters should be just as able to pick up from blue belts in 0.12 once the system is in equilibrium and supply is good enough.

I can't explain this in more ways. The only thing that matters is if input speed, output speed and transfer speed are good enough. Other than that there might be some wasted energy by the inserters when they fail to pick up items, but it doesn't affect production speed.