All the different ways to increase the efficiency of a nuclear power plant

[Khagan]

I suppose it depends on how often you either can't or won't build build enough fuel cell production for your reactors to decide if waiting for fuel is worth the added complexity. In my experience, never, so that's why I'm against it. Others might have a different experience.

There does seem to be a sharp split on this point. I agree that it is an unnecessary check. The thing you need to keep an eye on is the supply of U235: if you keep that up there should always be plenty of fuel cells, and you want to detect and fix any shortfall before it flows through to fuel production.

The other recurring disagreement is over the importance of the neighbour bonus. It seems to me that post-kovarex the main point of the bonus is not the saving on (cheap) fuel cells but the extra power produced by each (expensive) nuclear core. If you have designed your reactor to use the bonus, then not getting it leaves large amounts of expensive capital equipment being under utilised. Pre-kovarex the fuel saving is of course also important.

[foamy]

There does seem to be a sharp split on this point. I agree that it is an unnecessary check. The thing you need to keep an eye on is the supply of U235: if you keep that up there should always be plenty of fuel cells, and you want to detect and fix any shortfall before it flows through to fuel production.

Should.

I have had fuel cell production be interrupted because someone depowered the assembler, because the iron line got cut, because kovarex got cut, because a logistics chest got configured poorly, because someone deadlocked an intersection, etc. :v

[mmmPI]

I suppose it depends on how often you either can't or won't build build enough fuel cell production for your reactors to decide if waiting for fuel is worth the added complexity. In my experience, never, so that's why I'm against it. Others might have a different experience.

There does seem to be a sharp split on this point. I agree that it is an unnecessary check. The thing you need to keep an eye on is the supply of U235: if you keep that up there should always be plenty of fuel cells, and you want to detect and fix any shortfall before it flows through to fuel production.

To add to farcast conditions, i think it also depend on how the rest of the electric network look like, if you are using a mix of energy including solar and classic fuel burning, a shut-down of the nuke plant, due to not having enough fuel to keep the sync wouldn't lead to a total black-out, and may be your prefered behavior, if you follow this train of thought, then you can smartly use power switches to isolate the kovarex process to give it all available power, and/or science, to interrupt it while the fuel recover.

On the other side of the spectrum, if you have 300% nuclear capacity compared to consumption, but it's only a 2x2 or a 2x3 2x4 that is required for such excess, because you just built the thing and haven't put all the modules in the miners as you planned yet, and you have only this nuclear power plant as the sole and unique source of electricity because your starting coal that powered initial steam engine is depleted. You don't want the nuke plant to totally shut-down, waiting for more fuel to be efficient, wouldn't happen, since everything's would shut down.

That's a secondary condition relative to the fuel/ uranium stock : depending on the rest of the factory, you may or may not have a way to deal nicely with a shortage that's the main one. Only then if you have a plan for shortage like back-up steam engines or shutting down parts of the factory, it make sense to be willing to stop entirely the nuke plant to only work when full efficiency is possible.

I think it cannot be discarded on the argument that it's often not necessary, if you consider "all the different ways to increase the efficiency" as a synchronization system for fuel insertion that include a fuel check to prevent insertion if full fuel efficiency cannot be met is amongst the possible way to handle the situation and can be adapated to some situation a player encounter.

I have had fuel cell production be interrupted because someone depowered the assembler, because the iron line got cut, because kovarex got cut, because a logistics chest got configured poorly, because someone deadlocked an intersection, etc. :v

Rotated a belt while chatting and typing a message without the chat open.

Better safe than sorry, i can understand why one would want to monitor if the production level are adequate, and also if the delivery of the fuel is happening properly. That's 2 differents checks. the later can easily be embedded into the nuclear plant, contrary to the first one if you produce and consume in decentralised fashion.

[xaetral]

Similar problem, the pumps only manage 1200 steam/s so that's a bottleneck.

There are two (or one and a half) pumps per 16 heat exchangers, I know the pic is a quite zoomed out but they are there.

[Khagan]

I have had fuel cell production be interrupted because someone depowered the assembler, because the iron line got cut, because kovarex got cut, because a logistics chest got configured poorly, because someone deadlocked an intersection, etc. :v

As farcast said: 'different experience'.

• 'Someone' presumably means 'someone else'. I've not tried the joys(?) of MP, so I admit that guarding against casual vandalism is not on my priority list.
• Why is the fuel assembler more in danger of depowering than, say, the inserters that load the fuel into the reactors? Is that a possibility you want to guard against? (Hint: I supply them both from the same power loop.)
• A hiccup in iron supplies is going to be noticed in many other places long before it becomes significant for fuel cell production (which should be a high priority customer for the iron). And if you really are worried about it, monitor the iron supply to the assembler, not the stock of fuel cells.
• Kovarex can be temperamental, but that is covered by my recognition that supplies of U235 do need to be monitored.
• A configuration error is a construction error, and such things don't occur spontaneously; if you have one, fix your blueprint so it doesn't happen again. (In any case, I confine my use of logistic bots to construction supplies and munitions; there are none anywhere in my nuclear supply chain.)
• As with a shortage of iron, given how infrequently the fuel cell assembler actually needs resupply of raw materials, a deadlocked intersection is going to be noticed by its effects on other parts of the factory long before it interrupts fuel cell production.

[mrvn]

I suppose it depends on how often you either can't or won't build build enough fuel cell production for your reactors to decide if waiting for fuel is worth the added complexity. In my experience, never, so that's why I'm against it. Others might have a different experience.

There does seem to be a sharp split on this point. I agree that it is an unnecessary check. The thing you need to keep an eye on is the supply of U235: if you keep that up there should always be plenty of fuel cells, and you want to detect and fix any shortfall before it flows through to fuel production.

The other recurring disagreement is over the importance of the neighbour bonus. It seems to me that post-kovarex the main point of the bonus is not the saving on (cheap) fuel cells but the extra power produced by each (expensive) nuclear core. If you have designed your reactor to use the bonus, then not getting it leaves large amounts of expensive capital equipment being under utilised. Pre-kovarex the fuel saving is of course also important.

We are talking about efficient reactors so a design for neighbor bonuses is a given.

[mrvn]

I have had fuel cell production be interrupted because someone depowered the assembler, because the iron line got cut, because kovarex got cut, because a logistics chest got configured poorly, because someone deadlocked an intersection, etc. :v

As farcast said: 'different experience'.

• 'Someone' presumably means 'someone else'. I've not tried the joys(?) of MP, so I admit that guarding against casual vandalism is not on my priority list.
• Why is the fuel assembler more in danger of depowering than, say, the inserters that load the fuel into the reactors? Is that a possibility you want to guard against? (Hint: I supply them both from the same power loop.)
• A hiccup in iron supplies is going to be noticed in many other places long before it becomes significant for fuel cell production (which should be a high priority customer for the iron). And if you really are worried about it, monitor the iron supply to the assembler, not the stock of fuel cells.
• Kovarex can be temperamental, but that is covered by my recognition that supplies of U235 do need to be monitored.
• A configuration error is a construction error, and such things don't occur spontaneously; if you have one, fix your blueprint so it doesn't happen again. (In any case, I confine my use of logistic bots to construction supplies and munitions; there are none anywhere in my nuclear supply chain.)
• As with a shortage of iron, given how infrequently the fuel cell assembler actually needs resupply of raw materials, a deadlocked intersection is going to be noticed by its effects on other parts of the factory long before it interrupts fuel cell production.

All of that is still missing half the point of checking the fuel. The part about getting a fuel starved reactor back to running efficiently.

Seriously, checking fuel means adding a few wires. Depending on your control logic you don't even need a single extra combinator. If you never run into a low fuel situation then checking for fuel or not is irrelevant. But at startup, right after you blueprinted the reactor and fuel starts coming in it is helpful. So why skip it?

With fuel checking you can build a 12 cores reactor with one covarex centrifuge and assorted other fuel production stuff and make it self starting and put it all into a single blueprint. Then whenever you need more power you just blueprint and forget. Without the fuel check you have to manually insert some fuel so the reactor gets into the efficient state.

[gGeorg]

I am sorry, I have to finish our talk. Your arguments are nonsense and are piling up.

[foamy]

I have had fuel cell production be interrupted because someone depowered the assembler, because the iron line got cut, because kovarex got cut, because a logistics chest got configured poorly, because someone deadlocked an intersection, etc. :v

As farcast said: 'different experience'.

• 'Someone' presumably means 'someone else'. I've not tried the joys(?) of MP, so I admit that guarding against casual vandalism is not on my priority list.
• Why is the fuel assembler more in danger of depowering than, say, the inserters that load the fuel into the reactors? Is that a possibility you want to guard against? (Hint: I supply them both from the same power loop.)
• A hiccup in iron supplies is going to be noticed in many other places long before it becomes significant for fuel cell production (which should be a high priority customer for the iron). And if you really are worried about it, monitor the iron supply to the assembler, not the stock of fuel cells.
• Kovarex can be temperamental, but that is covered by my recognition that supplies of U235 do need to be monitored.
• A configuration error is a construction error, and such things don't occur spontaneously; if you have one, fix your blueprint so it doesn't happen again. (In any case, I confine my use of logistic bots to construction supplies and munitions; there are none anywhere in my nuclear supply chain.)
• As with a shortage of iron, given how infrequently the fuel cell assembler actually needs resupply of raw materials, a deadlocked intersection is going to be noticed by its effects on other parts of the factory long before it interrupts fuel cell production.

Eh, not vandalism, just refactoring that didn't check everything, coupled with production units that may or may not all have been designed by the same person. Sometimes I've been the one guilty of it, sometimes it's been another person. Stuff happens.

Careful separation of power grids works fine until someone puts a power pole in the wrong place, something that I have seen hard-crash some reactor designs by accidentally connecting their inner and outer power grids, and then someone goes and builds a bunch of chip farms or something and overloads the whole system

I don't understand the resistance to adding some extra wires, particularly since you're taking other steps to ensure you have fuel cells. Why not monitor it so that you know?

You want the chests anyway so that fuel insertion happens as rapidly as possible when triggered and doesn't try to pull from a moving belt (and therefore introduce potential desyncs), so it doesn't increase the footprint notably even for teeny tiny 2x2 reactors, let alone a full on 2xN machine.


(Oh, and in case people are misunderstanding me: MP is a blast and I love it. It's very interesting seeing people's visions collide and/or synergize into something better.)

[Khagan]

To add to farcast conditions, i think it also depend on how the rest of the electric network look like, if you are using a mix of energy including solar and classic fuel burning, a shut-down of the nuke plant, due to not having enough fuel to keep the sync wouldn't lead to a total black-out, and may be your prefered behavior, if you follow this train of thought, then you can smartly use power switches to isolate the kovarex process to give it all available power, and/or science, to interrupt it while the fuel recover.

On the other side of the spectrum, if you have 300% nuclear capacity compared to consumption, but it's only a 2x2 or a 2x3 2x4 that is required for such excess, because you just built the thing and haven't put all the modules in the miners as you planned yet, and you have only this nuclear power plant as the sole and unique source of electricity because your starting coal that powered initial steam engine is depleted. You don't want the nuke plant to totally shut-down, waiting for more fuel to be efficient, wouldn't happen, since everything's would shut down.

Good point. I'm definitely in the latter category anywhere before the multimegafactory scale. Until then, each time I build a nuclear plant, it multiplies my total power production capacity by a factor of 4 or 5.

[W]henever you need more power you just blueprint and forget. Without the fuel check you have to manually insert some fuel so the reactor gets into the efficient state.

Except of course that 'blueprint and forget' is actually

1. Place blueprint for power plant.
2. Place blueprint for rail station and connection to rail network.
3. Ensure work gang of construction bots is on site with roboport coverage.
4. Arrange delivery of raw materials to site.
5. Connect water supplies.

Manually inserting one fuel cell is trivial compared to any of the above.

You can avoid (5) by building on water, but then 'Place blueprint' becomes 'Place blueprint; wait for landfill to be constructed; place blueprint again'. And once you are building on water the cost-benefit balance swings in favour of 'always on' running (no steam tanks for controlling consumption), which probably doesn't belong in this thread.

You want the chests anyway so that fuel insertion happens as rapidly as possible when triggered and doesn't try to pull from a moving belt (and therefore introduce potential desyncs)

The belt has had at least 200s since the last trigger to return to a backed-up state. It isn't moving.

[foamy]

The belt has had at least 200s since the last trigger to return to a backed-up state. It isn't moving.

It shouldn't be moving. Something as simple as a reactor expansion can change that.

You could, I suppose, just monitor the belt overall for when it is in a backed up state. It'd more or less be the same check and it could be done with a fixed amount of wiring. Would require combinators, though.

[W]henever you need more power you just blueprint and forget. Without the fuel check you have to manually insert some fuel so the reactor gets into the efficient state.

Except of course that 'blueprint and forget' is actually

1. Place blueprint for power plant.
2. Place blueprint for rail station and connection to rail network.
3. Ensure work gang of construction bots is on site with roboport coverage.
4. Arrange delivery of raw materials to site.
5. Connect water supplies.

Manually inserting one fuel cell is trivial compared to any of the above.

You don't need a fuel check necessarily, but if you're using the output inserter as a memory-less clock it doesn't fail to an automatic load if, for whatever reason, fuel is interrupted. If you have no fuel cell available and the output inserter tries to trigger the input inserter, the input insertion will fail and not recover until the player intervenes and manually handloads either a fresh or spent fuel cell into the reactor.

Autoload designs avoid this.

[farcast]

One covarex centrifuge can produce fuel for 13 reactors. Assuming you don't use any of it for nukes or nuclear fuel for trains. I don't know how many centrifuge per reactor you need to process uranium ore before covarex but it's certainly way more than 1 per 60 reactors. And you also need a ton of uranium ore, meaning many miners and a lot of sulfuric acid. The chain of things needed to produce nuclear fuel isn't just the centrifuge and assembler to make the final product.

But lets say you do have covarex researched and setup one centrifuge to run it. And you build a 12 reactors setup. Because 13 > 12, plenty of over production of fuel cells. So you are golden, right?

Wrong. If fuel production is ever interrupted, e.g. the aliens destroyed a power pole at the wrong place, your reactor would need 2 covarex centrifuges to recover. With fuel checking it gets back to full power on it's own.

A reactor can't consume fuel faster than when it's running at full heat output. If one centrifuge is enough to run 12 reactors at full power then one centrifuge is enough to recover from any stage of brownout, as long as you can bring it back to running at full speed (or fast enough to meet the fuel demand for full power output).

The way I see it, waiting for fuel is only helpful in the specific situation where fuel production is not enough to run the reactors at full power. In that case, it prevents brownouts caused by fuel shortage from getting worse as quickly, and will always recover to the maximum power output your fuel production can theoretically provide. If there's enough fuel production to run reactors at full power, even if it's just enough, then waiting for fuel isn't necessary to return to max efficiency.

[Khagan]

If you have no fuel cell available and the output inserter tries to trigger the input inserter, the input insertion will fail and not recover until the player intervenes and manually handloads either a fresh or spent fuel cell into the reactor.

To get to that point, you need to leave the power station running steadily with no fuel being added to the belt for about two and a half hours. If nothing else, the steady decline in power output should have been noticed long before then.

[mrvn]

One covarex centrifuge can produce fuel for 13 reactors. Assuming you don't use any of it for nukes or nuclear fuel for trains. I don't know how many centrifuge per reactor you need to process uranium ore before covarex but it's certainly way more than 1 per 60 reactors. And you also need a ton of uranium ore, meaning many miners and a lot of sulfuric acid. The chain of things needed to produce nuclear fuel isn't just the centrifuge and assembler to make the final product.

But lets say you do have covarex researched and setup one centrifuge to run it. And you build a 12 reactors setup. Because 13 > 12, plenty of over production of fuel cells. So you are golden, right?

Wrong. If fuel production is ever interrupted, e.g. the aliens destroyed a power pole at the wrong place, your reactor would need 2 covarex centrifuges to recover. With fuel checking it gets back to full power on it's own.

A reactor can't consume fuel faster than when it's running at full heat output. If one centrifuge is enough to run 12 reactors at full power then one centrifuge is enough to recover from any stage of brownout, as long as you can bring it back to running at full speed (or fast enough to meet the fuel demand for full power output).

The way I see it, waiting for fuel is only helpful in the specific situation where fuel production is not enough to run the reactors at full power. In that case, it prevents brownouts caused by fuel shortage from getting worse as quickly, and will always recover to the maximum power output your fuel production can theoretically provide. If there's enough fuel production to run reactors at full power, even if it's just enough, then waiting for fuel isn't necessary to return to max efficiency.

Ok, got me there. Bad example. In that scenario the problem will be the miners. Because while it's easy to put the centrifuge on an internal power grid with the reactor in the blueprint the uranium miners are somewhere far far away and probably not on a separate power grid.

The centrifuges are only a problem if they are scaled to the power consumption and not the max fuel consumption of the reactor.

[foamy]

If you have no fuel cell available and the output inserter tries to trigger the input inserter, the input insertion will fail and not recover until the player intervenes and manually handloads either a fresh or spent fuel cell into the reactor.

To get to that point, you need to leave the power station running steadily with no fuel being added to the belt for about two and a half hours. If nothing else, the steady decline in power output should have been noticed long before then.

All it requires is a single gap, which can happen e.g. when expanding a reactor, or if you were using a bot delivery system and things got borked, or whatever.

[mmmPI]

I don't understand the resistance to adding some extra wires, particularly since you're taking other steps to ensure you have fuel cells. Why not monitor it so that you know?

Seriously, checking fuel means adding a few wires. Depending on your control logic you don't even need a single extra combinator. If you never run into a low fuel situation then checking for fuel or not is irrelevant. But at startup, right after you blueprinted the reactor and fuel starts coming in it is helpful. So why skip it?

Knowing that fuel is ready to be burn is nice, but if you design a fuel check that auto-prevent loading when it isn't, in some situations, like having only 1 extra-sized nuclear plant and nothing else to back it up, it could transform a "potential" spiraling brown-out to an instant unrecoverable black-out.

This could be caused by just 1 massive spike in energy consumption, like a new artillerie area, or the deconstruction of an unsually large forest. From a period of say 5 to 10 minutes you may be having a fuel shortage due to your self-regulated system consuming like crazy, and the production of your centrifuge that was "countperfect" being a little slowed due to the spike in electricity consumption. If you stop your system to wait for more fuel let say 1 second before the refuel should have happened you could be ruined.

It could feel like introducing a safety-mechanism, whose potential failure is the only hazard.

If it only tells you with an alarm it's different.

Or if you can automate an answer.

Except of course that 'blueprint and forget' is actually

Place blueprint for power plant.
Place blueprint for rail station and connection to rail network.
Ensure work gang of construction bots is on site with roboport coverage.
Arrange delivery of raw materials to site.
Connect water supplies.

Manually inserting one fuel cell is trivial compared to any of the above.

i think that too, because generally i build 1 or 2 maybe 3 nuclear plant, the starting one when i can, then a way larger one that i think will cover the need for the rest of the game ( i tend to have 200 250H hours games in average) then sometimes a 3rd one when the 2nd wasn't big enough. The time investment isn't much.

On the other hand, play a grid-like map, where you copy paste small nuclear plant like it's solar pannel, you design your thing so that when you paste it it fit the train station that is already there, covered by construction bots, all material already there and water supply aligned , or even coming by train ! Which nullify, i think in this particular case, all objections

If you have no fuel cell available and the output inserter tries to trigger the input inserter, the input insertion will fail and not recover until the player intervenes and manually handloads either a fresh or spent fuel cell into the reactor.

To get to that point, you need to leave the power station running steadily with no fuel being added to the belt for about two and a half hours. If nothing else, the steady decline in power output should have been noticed long before then.

All it requires is a single gap, which can happen e.g. when expanding a reactor, or if you were using a bot delivery system and things got borked, or whatever.

True, true, also true .

You don't need a fuel check necessarily, but if you're using the output inserter as a memory-less clock it doesn't fail to an automatic load if, for whatever reason, fuel is interrupted. If you have no fuel cell available and the output inserter tries to trigger the input inserter, the input insertion will fail and not recover until the player intervenes and manually handloads either a fresh or spent fuel cell into the reactor.
Autoload designs avoid this.

When i did the nuclear plant, with output inserter clocked at 200sec + X%, triggering the refuel, i used belt to deliver fuel, noticed the risk of gaps you mention, and made sure the belt was always loaded from some chest wired to an alarm to monitor the buffer. It's halfway between monitoring the fuel in chest and monitoring the U235 as Khagan says. The fact that the inserter where synced via clock meant that none of them could create a gap for another one on the belt. The startup phase was me putting manually a fuel cell in each reactor, then putting power so they are all consumed at the same exact time. First cycle was also used to test the buffering capacity. Then 300 sec later, first ejection of all used fuel, and things were synced and stayed. Definitly not copy-paste and forget. I was looking at the whole process haha, checking each heat pipes , seeing the temperature, not willing to have any area reach 1000°.

It wasn't a MP game and it was heavily moded. I avoided messing around the nuclear reactor building after it was made, but i happily played with the slider to make it burn fuel faster and faster as the game progress, knowing that i would be told by my alarm if my different areas sending fuel couldn't meet the demand, the buffer of fuel giving me time to react manually, ie: making more centrifuges! And when i reached 200sec + 10% i knew i had to build a larger one

You need to be aware of energy consumption at all time which is often impossible on 24/24 server on which prefered designs are more robusts imo.

[mrvn]

I don't understand the resistance to adding some extra wires, particularly since you're taking other steps to ensure you have fuel cells. Why not monitor it so that you know?

Seriously, checking fuel means adding a few wires. Depending on your control logic you don't even need a single extra combinator. If you never run into a low fuel situation then checking for fuel or not is irrelevant. But at startup, right after you blueprinted the reactor and fuel starts coming in it is helpful. So why skip it?

Knowing that fuel is ready to be burn is nice, but if you design a fuel check that auto-prevent loading when it isn't, in some situations, like having only 1 extra-sized nuclear plant and nothing else to back it up, it could transform a "potential" spiraling brown-out to an instant unrecoverable black-out.

This could be caused by just 1 massive spike in energy consumption, like a new artillerie area, or the deconstruction of an unsually large forest. From a period of say 5 to 10 minutes you may be having a fuel shortage due to your self-regulated system consuming like crazy, and the production of your centrifuge that was "countperfect" being a little slowed due to the spike in electricity consumption. If you stop your system to wait for more fuel let say 1 second before the refuel should have happened you could be ruined.

Or you could be saved. Because during that one second all your science production will slow, fuel production slows too but continues, lasers have priority so they still go full blast. And then fuel is there and the reactor goes full blast again instead of running at a fraction of power and never getting back to full power.

There are 2 dangers here: running low on power for a short time and running low on power until you manually fix it. I prefer the former but then again I never only have one source of power that is a single point of failure.

It could feel like introducing a safety-mechanism, whose potential failure is the only hazard.

If it only tells you with an alarm it's different.

Or if you can automate an answer.

Making it only sound an alarm is an option. You can also always have a chest with some emergency fuel somewhere.

Note: The fuel check in my designs, or any design that uses the removal of the spend fuel cell as trigger, is necessary. If you ever remove a spend fuel cell without a fresh fuel cell to put into the reactor the setup dies. That's really the only argument I need.

Except of course that 'blueprint and forget' is actually

Place blueprint for power plant.
Place blueprint for rail station and connection to rail network.
Ensure work gang of construction bots is on site with roboport coverage.
Arrange delivery of raw materials to site.
Connect water supplies.

Manually inserting one fuel cell is trivial compared to any of the above.

• Place blueprint with power poles, roboports and radars with landfill under them.
• Repeat till they are actually all build and you have complete construction coverage.
• Place landfill blueprint for the reactor.
• Place blueprint for the reactor.

If you don't have a global construction zone then you would start with the rail station. You would have to build that manually. Arranging delivery of raw materials would also be manual. But after that it's just placing the blueprints. You can do that via map then. I really wish one could place a blueprint with landfill and entities in one go. I should check if one could mod that so a ghost entity can be placed on top of a ghost tile.

If you have no fuel cell available and the output inserter tries to trigger the input inserter, the input insertion will fail and not recover until the player intervenes and manually handloads either a fresh or spent fuel cell into the reactor.

To get to that point, you need to leave the power station running steadily with no fuel being added to the belt for about two and a half hours. If nothing else, the steady decline in power output should have been noticed long before then.

All it requires is a single gap, which can happen e.g. when expanding a reactor, or if you were using a bot delivery system and things got borked, or whatever.

I don't really see that as a problem. That's what the chest before every reactor is for. After refueling you have 200s time for the inserter to pick up a fuel cell from the belt and put it into the chest. And then fuel can be inserted into the reactors from the chest with a single tick pulse. No need for a longer pulse to handle the delay of inserters picking fuel from the belt.

True, true, also true .

You don't need a fuel check necessarily, but if you're using the output inserter as a memory-less clock it doesn't fail to an automatic load if, for whatever reason, fuel is interrupted. If you have no fuel cell available and the output inserter tries to trigger the input inserter, the input insertion will fail and not recover until the player intervenes and manually handloads either a fresh or spent fuel cell into the reactor.
Autoload designs avoid this.

When i did the nuclear plant, with output inserter clocked at 200sec + X%, triggering the refuel, i used belt to deliver fuel, noticed the risk of gaps you mention, and made sure the belt was always loaded from some chest wired to an alarm to monitor the buffer. It's halfway between monitoring the fuel in chest and monitoring the U235 as Khagan says. The fact that the inserter where synced via clock meant that none of them could create a gap for another one on the belt. The startup phase was me putting manually a fuel cell in each reactor, then putting power so they are all consumed at the same exact time. First cycle was also used to test the buffering capacity. Then 300 sec later, first ejection of all used fuel, and things were synced and stayed. Definitly not copy-paste and forget. I was looking at the whole process haha, checking each heat pipes , seeing the temperature, not willing to have any area reach 1000°.

It wasn't a MP game and it was heavily moded. I avoided messing around the nuclear reactor building after it was made, but i happily played with the slider to make it burn fuel faster and faster as the game progress, knowing that i would be told by my alarm if my different areas sending fuel couldn't meet the demand, the buffer of fuel giving me time to react manually, ie: making more centrifuges! And when i reached 200sec + 10% i knew i had to build a larger one

You need to be aware of energy consumption at all time which is often impossible on 24/24 server on which prefered designs are more robusts imo.

That's a lot of fuel cells to produce to fill the belt before the reactor can run safely. I usually start producing U238/U235 early and just buffer it till I can start covarex. By the time the research for that is done there should be enough U238/U235 to get a centrifuge running. While that spins up and actually starts producing fuel I build a simple 2x2 reactor. If I then have to wait for 1000 fuel cells to be produced to fill the belt before turning on the reactor I would be pissed. I want it started as soon as 4 fuel cells are ready.


Measuring the time between refuels to raise an alarm when it's time to build the next reactor is nice. But only tells you about the average energy consumption. Not even that exactly. There could be lots of spikes in there exceeding the maximum output where you have brownouts. Measuring the steam throughput or an accumulator is a better detector as it will tell you when spikes exceed your maximum power. And measuring fuel availability tells you when you need more centrifuges. Measuring uranium tells you when you need more ores, and so on. You can set an alarm at every step of the process if you really want.

[mmmPI]

Or you could be saved. Because during that one second all your science production will slow, fuel production slows too but continues, lasers have priority so they still go full blast. And then fuel is there and the reactor goes full blast again instead of running at a fraction of power and never getting back to full power.

Well i wanted to explicitly point out the worst case possible, where the temperature of your whole system is 501°, and there is almost no steam buffer. It could be the case if you had been under heavy load for 10 minutes straight, the refuel in this case cannot be delayed for say 45 sec because the last fuel required to reach full efficiency has been delayed due to brown out. In this case if you attempt to wait for the fuel to arrive, you will have loss all power when it does and can't insert anything.

The risk of delaying refuel isn't always synonym of instant nuke plant shut down, but on 2x4, like pretty small, it could.

Note: The fuel check in my designs, or any design that uses the removal of the spend fuel cell as trigger, is necessary. If you ever remove a spend fuel cell without a fresh fuel cell to put into the reactor the setup dies. That's really the only argument I need.

that's a better way to put it, i agree.

another way to do it could be to check only if the last injecting inserter has some fuel, first fuel on the belt will be grabbed by the first inserter, but it will be unable to finish the swing due to the used fuel cell present in the core. Then second fuel on the belt will get grabbed by the second inserter and so on. It requires initialiazing the plant with used fuel cell instead of fresh one though. I had thought about it, it doesn't require loads of buffer as you mention.

That's a lot of fuel cells to produce to fill the belt before the reactor can run safely. I usually start producing U238/U235 early and just buffer it till I can start covarex. By the time the research for that is done there should be enough U238/U235 to get a centrifuge running. While that spins up and actually starts producing fuel I build a simple 2x2 reactor. If I then have to wait for 1000 fuel cells to be produced to fill the belt before turning on the reactor I would be pissed. I want it started as soon as 4 fuel cells are ready.

It was a lot, but i was designing the nuke plant, not using an existing blueprint and i was slow at it compared to the fuel cell production that was buffered, that happens a lot when i play with different ruleset/mods

I used only 1 side of the belt for fresh fuel, the other for used fuel. On the plus side, the quantity of fuel on the belt act as buffer, when the train chest runs dry the alarm ring, the time the belt empty is giving me time to react. In case it had failed, i would have had to run on site, take fuel on the belt , put 5 of them in each core, panic mode, quickly add centrifuge and uranium miner, and sulfuric acid, oh damn no more oil, that will take an hour to solve I'M GLAD I HAVE 1000 FUEL CELL ON THE BELT!

Also modded games can make correcting the supply chain a very long thing ! I know buffer are not popular but i like them.

Measuring the time between refuels to raise an alarm when it's time to build the next reactor is nice. But only tells you about the average energy consumption. Not even that exactly. There could be lots of spikes in there exceeding the maximum output where you have brownouts. Measuring the steam throughput or an accumulator is a better detector as it will tell you when spikes exceed your maximum power. And measuring fuel availability tells you when you need more centrifuges. Measuring uranium tells you when you need more ores, and so on. You can set an alarm at every step of the process if you really want.

i didn't smart deep enough to go measuring in game the time between refuel and automate things, i knew the theoric capacity was around 2GW i think, then i looked occasionally on the power graph my electric consumption, and set up the refuel time accordingly. Like 500MW i would go for roughly 30% capacity to give me margin.( refuel every 200sec+70%) Later i added row of accumulators to smooth out the laser tower consumption dealing with those spike you mention, i didn't used bots until way later when i added solar pannel over time when i had access to more surface. I then increasingly capped the nuclear plant to save my uranium to make bombs to get more surface. or maybe it was to make plutonium, maybe both.

I added an alarm on the accumulators at 80% or something. It was not ringing when laser tower had actions, only in case i forgot to increase capacity on the nuke plant. Only 2 alarm, one that check if fuel was delivered properly and 1 if brown-out was menacing. One case, i need to add to production chain, other case i need to decrease refuel time.Too many alarms is annoying

[gGeorg]

You don't need a fuel check necessarily, but if you're using the output inserter as a memory-less clock it doesn't fail to an automatic load if, for whatever reason, fuel is interrupted. If you have no fuel cell available and the output inserter tries to trigger the input inserter, the input insertion will fail and not recover until the player intervenes and manually handloads either a fresh or spent fuel cell into the reactor.

Autoload designs avoid this.

add a memory. So the signal "used_cell_out" is saved into memory, then memory is reset on "fresh_cell_in".
As result, in case of fuel shortage, loading inserter waits as long as needed.
Not sure if you mean "Autoload desing" equals a memory cell.

[foamy]

Note: The fuel check in my designs, or any design that uses the removal of the spend fuel cell as trigger, is necessary. If you ever remove a spend fuel cell without a fresh fuel cell to put into the reactor the setup dies. That's really the only argument I need.

This issue is specific to memory-less designs (i.e.the combinator-free one).

You just need a simple memory cell and you can use a spent-cell trigger without that problem. As long as the logic is set up properly, it will start in 'I need fuel' and hold that state until a fresh cell is inserted, at which point it holds 'I don't need fuel' until the spent cell is removed.

Takes one combinator, patches the immediate hard failure on fuel interruption problem.