Nuclear Reactors BluePrint Cloning

[Tekillaa]

TL;DR

Option to calculate Nuclear Reactor Electric Central BP Clone like the game do for solar panel and accumulator.

What ?

Is it possible to make an option(or maybe a mod?) that can identify a clone of the same "nuclear reactor blueprint" to calculate all others iterations of the blueprint as clone of the first? A bit like the game do for solar panel? It can be good to use lot of nuclear reactors with a moderate cost for updates efficiency?
List of conditions to be counted as clone: (?)(Just my first sight, maybe need few more conditions?)
- Close fluid loop : water supply include (water pump in), no fluid come in or out from the BP (water and steam)
- Perfect feeding for reactors : reactors have to be filled with nuclear fuel cells and have already consumed one or 2 so the BP is already "warmed up"
- Same electric network.
With a deeper reflexion, I think it's possible to extend the way of thinking for a 2n reactor too.
Edit : For the player : show somehow what are the "clones" (thks SoShootMe)

Why ?

To make nuclear energy viable for ultra late game/megafactory!
Usually people use 1 to 3 Blueprint of a same nuclear central: most of the time, waterpumps are in it thanks to landfilling, and the point of this way of thinking is to make a moderate cost for updates efficiency with electricity coming from nuclear reactors : the game process only one BP and others as clone of the first as long few conditions are filled and people will be able to create their own "power plant" BP, so it will not limit the player creativity. Like a marker announcing the bp as "power plant"

I know this idea is wierd, but I feel the game needs it to put nuclear energy as viable option for megafactory to not have only solar panels/accumulators fields as solution.

thanks, have fun, take care!

[ssilk]

Reminds me to viewtopic.php?f=6&t=82060 Optimization idea: abstraction

[Tekillaa]

Exactly, but the conditions to be able to consider a "factory module" as clone is far more complicated considering the many inputs and outputs a factory can have. But for a powerplant, i feel it's really easy to make it work nice and the game will earn a good performance optimisation.

[SoShootMe]

Reminds me to viewtopic.php?f=6&t=82060 Optimization idea: abstraction

That was my first thought too, but the concept is a bit different. For "abstraction" the idea amounts to somehow validating a set of things can be replaced with a simpler model, then using that model instead of the full simulation. With "cloning" the idea is to have the full simulation still, but only for a single instance, and then duplicate the externally visible effects.

In both cases, I can only see it being remotely practical with boundaries defined by the player, eg something like Factorissimo. And if that is the case, I would say it is a fairly major change to the current gameplay: by itself probably enough to say it's extremely unlikely to happen, without considering other challenges (of which I think many remain, eg cloning would effectively require instances to be synchronised).

At least, I think there is more chance of fluid simulation changes improving performance sufficiently to make nuclear power viable for megabases, something that has broader benefit with less (or no) effect on gameplay.

[ssilk]

Ok, let it repeat me as so:
This is about copying a part of a factory many times. And given, that this part has the same input as the first, it should produce the same output.

Hm. I think the idea is cool but it is not so much easier, than the other. I think they have nearly the same complexity. Nevertheless this should be tried out somehow (as the abstraction idea).

[Tekillaa]

Dave McW found me a mod very near from the idea I suggest :
https://mods.factorio.com/mod/some-npp
but it's not completly my idea yet : I wish people to be able to build their own powerplant.

A nuclear powerplant with a BP with waterpump in got only one input : uranium fuel cells, and it's easy to create a condition checking if all reactor are filled and warmed up, making all clone as the same status as the first construction. Output being electricity on a same network and used up uranium cells, the "synchronisation" I ask are mostly for what goes beetween reactor and electricity output (heatpipes, pipes, heat exchanger and turbine) to avoid the same heat and fluid calculation for each powerplant if they are the same BP.

Abstraction is a way but this is not my first suggestion.

I would say it is a fairly major change to the current gameplay

In my POV, it's not entirely true : It will change nothing for the game play, people are still free to use as many different power plant BP they want, but there is at least a way to go big if necessary without the computer making calculation for something that we can consider as the same.

It's about a powerplant only, not a factory. This way to think works only because powerplants are on the same electricity network, they all share the same output. It's because of that condition that I believe my suggestion will work fine.

[SoShootMe]

I would say it is a fairly major change to the current gameplay

In my POV, it's not entirely true : It will change nothing for the game play, people are still free to use as many different power plant BP they want, but there is at least a way to go big if necessary without the computer making calculation for something that we can consider as the same.

You quoted without context but I was saying it seemed something like Factorissimo warehouses would be necessary to provide a means for the player to define the boundaries of what may be cloned. What I had in mind as "a fairly major change" was the space reduction, but I think it applies anyway, as there would have to be some other benefit to "boxing" the thing to be cloned, otherwise it's just... weird. Either way the result is a change how people play - though not necessarily how they must play - which is all I meant.

It's about a powerplant only, not a factory. This way to think works only because powerplants are on the same electricity network, they all share the same output. It's because of that condition that I believe my suggestion will work fine.

Being on the same electric network is probably necessary, but as I see it there's nothing really special about cloning power plants (solar aside) compared to something else. For a nuclear power plant, there's the input of uranium fuel cells and water, and the output of used up uranium fuel cells; also, some energy is produced. A battery factory might have input of water, oil and ores, and output of batteries; also, some energy is consumed.

[Tekillaa]

Ok, sorry for my missunderstanding.

I was suggesting this set of conditions to be counted as clone :

- Close fluid loop : water supply include (water pump in), no fluid come in or out from the BP (water and steam)
- Perfect feeding for reactors : reactors have to be filled with nuclear fuel cells and have already consumed one or 2 so the BP is already "warmed up"
- Same electric network.

About the water, if waterpumps are not in, maybe water tanks levels as input conditions?
I'm not suggesting something about space reduction, it's about how the computer updates entities for a power plant : the goal is to reduce the number of entites to refresh each tick if we use Nuclear reactor. And for the "boundaries of what can be cloned" as long the set of conditions are filled, just an attribute on the powerplant BP? (something that make a link for entities such as: heat pipes, pipes, tanks, heat exchangers, turbines and pumps)
I was making that suggestion because most of megafactories use solarpanels/accumulator to improve the UPS efficiency and I think the game will earn quality if we let more solutions availables for players about energy production. (for 5 GW+).
If I limit my suggestion to a powerplant only (preferably with closed fluid system), it's because for a factory, too many things are to consider : water oils and ores? If inputs are not perfectly filled, how to reproduce each gap on a belt, so each inserter behavior? There is too many checks to do to apply this way of thinking to a factory, but it's trivial for a power plant as long reactors got fuels cells in.

[SoShootMe]

Ok, sorry for my missunderstanding.

Looks like I owe you a bigger apology...

I was suggesting this set of conditions to be counted as clone :

- Close fluid loop : water supply include (water pump in), no fluid come in or out from the BP (water and steam)
- Perfect feeding for reactors : reactors have to be filled with nuclear fuel cells and have already consumed one or 2 so the BP is already "warmed up"
- Same electric network.

Sorry, I should have addressed these conditions directly. I'm not sure how water pumps inside would work without strange restrictions to take advantage of cloning, and more importantly it doesn't address other inputs or outputs, but...

About the water, if waterpumps are not in, maybe water tanks levels as input conditions?

Yes, something like that could (theoretically) work. The same with chests for item inputs, which is a generalisation of reactors being filed with uranium fuel cells. Also similar for fluid and item outputs (but space for them instead of supply of them).

And for the "boundaries of what can be cloned" as long the set of conditions are filled, just an attribute on the powerplant BP?

As far as I know, once placed, there is no longer any connection to a blueprint, just a collection of things in the world. The attribute needs to persist, and I think the player needs to be able to see it. Something like a Factorissimo warehouse .

If I limit my suggestion to a powerplant only (preferably with closed fluid system), it's because for a factory, too many things are to consider : water oils and ores? If inputs are not perfectly filled, how to reproduce each gap on a belt, so each inserter behavior? There is too many checks to do to apply this way of thinking to a factory, but it's trivial for a power plant as long reactors got fuels cells in.

Too many checks is not really the problem, if they are only on the input and output; still far less than all the parts, that would be repeated many times over due to clones.

(I'll read any reply, but I suspect this has at least reached the point I have nothing more to contribute.)

[Tekillaa]

The attribute needs to persist, and I think the player needs to be able to see it. Something like a Factorissimo warehouse .

This is definitely what I was missing and it is essential, thanks a lot for pointing this out.

[ssilk]

Ok, let it repeat me as so:
This is about copying a part of a factory many times. And given, that this part has the same input as the first, it should produce the same output.

Hm. I think the idea is cool but it is not so much easier, than the other. I think they have nearly the same complexity. Nevertheless this should be tried out somehow (as the abstraction idea).

I thought a bit around this. I think meanwhile this idea has a great potential.
And it adds a really cool new game-element. And for testing purpose it should be possible to mod it.

Let’s say we use the Factorissimo2 mod for this.

Then this could be implemented by
- having the ability to copy a Factorissimo-factory and if you change something inside, it is changed in all copies.
- we need something to have the same state for all entities in all Factories. The simplest method would be a reset if a new factory is built. Which means, every item in all instances of this factory is then removed, the assemblies are reset, pipes/tanks are cleared, accumulators are empty, belts cleared, burner inserters cleared with one coal left. Etc. More advanced: some kind of synchronization takes place.
- every factory needs to be in the same electric network.
- items/fluids are only allowed to enter into that factory, if the same number of items are available at every factory. The same is with the output: it will only output something, if there is enough space left at every factory.
- technically, if something enters, it is produced only in the first factory. And only when it comes to the output the results are copied to the other factories.
- logically: when I enter any factory, I see always the same.

With all those preconditions, I think the simulation should really work identical.

Gameplay-wise this has great advantages over the abstraction-idea (viewtopic.php?f=6&t=82060 ). While with abstraction you don’t need to change anything, with cloning you need to built completely different. I see that as advantage.

[pichutarius]

last time i did a detailed experiment about heat transfer to find their formula, and guess what i found? adjacent nuclear plants heat transfer is calculated (transfered) 3 times per tick! Consuming all those precious ups. they probably do it to simulate the feeling that heat flow between plants is faster than heat pipes.

my guess is making connected nuclear plant group as a single entity will improve ups by a lot (they share the same temperature). also redo heat/fluid mechanics too.

[blazespinnaker]

items/fluids are only allowed to enter into that factory, if the same number of items are available at every factory. The same is with the output: it will only output something, if there is enough space left at every factory.

The problem with this solution is the lag time between input and output. Without having something that understands the innards, this won't necessarily be deterministic. I imagine there might be a way to store everything (data wise) for long periods, but it'll get time/space wise overly complex at scale.

Still, I suppose there are hacks you could do. You could periodically wipe the root factory of any and all production (including intermediate products) to reset your data storage requirements. That seems like a rather big sacrifice all for determinism and very very marginally greater accuracy.

Though, I suppose you could gamify it. For example, each time a child factory doesn't have enough resources it falls out and becomes inactive until a wipe of the root factory occurs. The player can set some value for each root factory of the ratio of child factories that fall out before a wipe occurs.

Not sure how much fun that would be in practice, though. It also seems rather orthogonal to a factory building game.

[ssilk]

last time i did a detailed experiment about heat transfer to find their formula, and guess what i found? adjacent nuclear plants heat transfer is calculated (transfered) 3 times per tick! Consuming all those precious ups.

How do you know, how this is been calculated?

they probably do it to simulate the feeling that heat flow between plants is faster than heat pipes.

Therefore it’s not needed to calculate it three time. Just another exchange factor between the plants should be enough.

my guess is making connected nuclear plant group as a single entity will improve ups by a lot (they share the same temperature).

No, they don’t. It depends very much of your layout how the temperature of the plants are spread. If you have a symmetric layout, you might be right.

The problem with this solution is the lag time between input and output.

Why?
I can think to different problems, but they are all solvable. For example there are small input/output-buffers, just the simplest solution of many.

Without having something that understands the innards, this won't necessarily be deterministic.

Why? If no input on any sub-factory nothing is input on all sub-factories. If output is clogged on any sub-factory nothing will be put out on all sub-factories.

This is fully deterministic and introduces a new gameplay: if you are overproducing too much, your production slows down.

I imagine there might be a way to store everything (data wise) for long periods, but it'll get time/space wise overly complex at scale.

dunno what you mean,

Still, I suppose there are hacks you could do. You could periodically wipe the root factory of any and all production (including intermediate products) to reset your data storage requirements. That seems like a rather big sacrifice all for determinism and very very marginally greater accuracy.

This Is never needed. Once the sub-factories synced, this will work fully deterministic. The game could calculate everything with one sub-factory and could take the results to create the rendering of all others. It would be no difference.

And would look quite interesting, not to say amazing.

[pichutarius]

...

Q: How do you know, how this is been calculated?
A: Lots of empirical data, come up with hypothesis formula, make prediction and test the result. (like real science)
I actually made a post 3 years ago but i guess it was too mathy and gain not much attention. but there was another guy that confirm my hypothesis.
viewtopic.php?f=5&t=50169 (beware, heavy math)

Q: Therefore it’s not needed to calculate it three time. Just another exchange factor between the plants should be enough.
A: The result is different. Just like other physic simulation, breaking into smaller steps gives smoother result, but cost more UPS. On the contrary, bigger exchange factor with bigger time steps save UPS but might get weird behavior, and i suspect that the result is more sensitive to build sequence. Dev can do some testing and see if the result is acceptable.

Q: No, they don’t. It depends very much of your layout how the temperature of the plants are spread. If you have a symmetric layout, you might be right.
I respectfully disagree. Imagine a 6-core nuclear plant blueprint. Now imagine those 6 were replaced with a HUGE single entity, eats fuel 6 times faster but generate 6 times more heat. All the heat transfer between the cores are gone. How much UPS save i dunno but i would say at least its not 0.

[blazespinnaker]

Still, I suppose there are hacks you could do. You could periodically wipe the root factory of any and all production (including intermediate products) to reset your data storage requirements. That seems like a rather big sacrifice all for determinism and very very marginally greater accuracy.

This Is never needed. Once the sub-factories synced, this will work fully deterministic. The game could calculate everything with one sub-factory and could take the results to create the rendering of all others. It would be no difference.

And would look quite interesting, not to say amazing.

Yes, been thinking more about this. Credit where credit is due, your idea is quite innovative.

But I'm not sure how well it will work in practice.

Imagine a complex factory for a moment with N assemblers, say something that takes raw ore / fluid and produces RCUs for example.

Here is a list of the intermediate products that first have to get built before
https://kirkmcdonald.github.io/calc.htm ... l-unit:f:1

https://wiki.factorio.com/Rocket_control_unit

According to the wiki, there is 98 seconds of assembly time before an RCU comes out the other side. Depending on the size of the factory, there could be a large mass of intermediate products.

This has a number of implications:

1. You have to wipe all the factories before syncing them.
2. You can't add/sync a new child factory unless you wipe the parent/child factories and reset them.
3. If a factory for whatever reason fails to have enough supply at any point, it is desynced and becomes inactive until there is a full wipe of factories.

It may work, but it will require that the player ensures there is enough buffer at all points for the factories to work. You'd want to leave a fair room for margin of error

If it ends up with factories constantly being desynced / wiped, might be a headache and a lot of sacrifice in the name of determinism, which may not be particularly relevant to someone at megabase scale.

The other issue is that this doesn't feel particularly authentic. It sort of feels like a shoehorned hack. I'm not sure what the real world simulation being done here is.

If you store the usage requirements as data as I mention above, this would allow for factories to produce at anytime as long as they can follow the usage path. If for some reason they don't have enough supply, they can just start over at the beginning once they have enough. But it will get expensive in terms of data and possibly processing requirements.

Still, at the very least it's worth a try. Your method is certainly very easy to implement. You could also set a parameter which allows for automatic wipe/syncing when a certain ratio of children are desynced.

You want the wipe code regardless of the approach you take to deal with user abuse / edge cases of hand feeding and filling chests inside factories.

[Yoyobuae]

Q: No, they don’t. It depends very much of your layout how the temperature of the plants are spread. If you have a symmetric layout, you might be right.
I respectfully disagree. Imagine a 6-core nuclear plant blueprint. Now imagine those 6 were replaced with a HUGE single entity, eats fuel 6 times faster but generate 6 times more heat. All the heat transfer between the cores are gone. How much UPS save i dunno but i would say at least its not 0.

From my testing the UPS cost of a nuclear powerplant is split as follows:

• Heat exchangers = nearly half of the UPS cost
• Steam turbines = nearly half of the UPS cost
• Heat pipes = the remainder UPS cost
• Nuclear reactors = negligible
• Pipes = negligible

This is assuming an actually UPS friendly design.

If you want to improve UPS of a nuclear powerplant using mods then target heat exchangers and steam turbines. A "big heat exchanger" and "big steam turbine" mod which replaces several of those entities by a single one would be a great start.

[pichutarius]

...

oowh, guess i stand corrected.

[Guenni7]

From my testing the UPS cost of a nuclear powerplant is split as follows:

• Heat exchangers = nearly half of the UPS cost
• Steam turbines = nearly half of the UPS cost
• Heat pipes = the remainder UPS cost
• Nuclear reactors = negligible
• Pipes = negligible

This is assuming an actually UPS friendly design.

If you want to improve UPS of a nuclear powerplant using mods then target heat exchangers and steam turbines. A "big heat exchanger" and "big steam turbine" mod which replaces several of those entities by a single one would be a great start.

Easiest way would be a Nuclear Reactor that needs fuel + water and directly gives out power instead of heat.

[ssilk]

pichutarius: You know, there is a quote-button?

Q: No, they don’t. It depends very much of your layout how the temperature of the plants are spread. If you have a symmetric layout, you might be right.
I respectfully disagree. Imagine a 6-core nuclear plant blueprint. Now imagine those 6 were replaced with a HUGE single entity, eats fuel 6 times faster but generate 6 times more heat. All the heat transfer between the cores are gone. How much UPS save i dunno but i would say at least its not 0.

Sorry for being not precise enough. You wrote:

guess is making connected nuclear plant group as a single entity will improve ups by a lot (they share the same temperature).

My answer was about the temperature. Not the ups . I mentioned that in my answer, but you seem to read over that.

1. You have to wipe all the factories before syncing them.

The easiest method is surely clearing all items and reset all entities.
But other methods are thinkable. The point is: it needs to be synchronized. It’s open how: More elegant methods are thinkable.

2. You can't add/sync a new child factory unless you wipe the parent/child factories and reset them.

Depends on, how that works in the end. Thinking to a mod (Factorissimo), that is surely the best and simplest method.

But when this is added to vanilla,I would say, there are separate and automated mechanisms for synchronization. I’m thinking to logistic robots and request. Like the modules requests. Adding a new child would mean then just to pause all child’s, until the new child is in synchronized state.
Besides that I would say, there should be always the option to clear everything, e.g. when you have no bots available or so.

3. If a factory for whatever reason fails to have enough supply at any point, it is desynced and becomes inactive until there is a full wipe of factories.

That could be one mode of operation. Yes.
But I already wrote: if there is not enough supply at one child, all child’s won’t work. They are fully synchronized.

It may work, but it will require that the player ensures there is enough buffer at all points for the factories to work. You'd want to leave a fair room for margin of error

The players task is to bring enough items to the entries and take the produced items out fast enough.

He hasn’t to think about errors, because there are none. Either everything is synchronized or not.

The other issue is that this doesn't feel particularly authentic. It sort of feels like a shoehorned hack. I'm not sure what the real world simulation being done here is.

This is much more authentic than the simulation idea, because it produces at every time exactly the same, as if the child’s are all separately simulated and not simulated by one parent.

PS: I will open an extra suggestion for this, I’m sorry break into this thread.