A Case for Balancing Nuclear Power

[Tesse11ation]

Introduction
So, we hear it all the time: "Solar power is waaaay better than nuclear." Why is that so? Nuclear power is simply not properly balanced in its current form. I'd like to first establish the main reasons for solar being better, why nuclear is not sustainable, and then give my proposals as to what might allow nuclear to compete mid- to late-game. Because I'm sure we all want to actually have fun with nuclear power.

What makes solar better?
Solar power is a strange beast, and it's a bit broken all on its own. Solar produces massive amounts of energy per tile to the point where you have to suspend your disbelief to use it. Yes, this is an acceptable break from reality since it's a video game, and not many people in Factorio find true joy in slaving away at placing hundreds of thousands of solar panels down by hand. So, it's not likely that nerfing solar would be a good solution.

What else makes solar energy so powerful? Updates per second. Late-game, solar has a negligible UPS cost for its size. This is great, but it also means that by late-game, nuclear simply is not viable by comparison. This has to be by far the number one reason why nobody... and I mean NOBODY... uses nuclear for their megabase.

Why does nuclear power suck?
Aside from the UPS issues, nuclear power is actually quite underwhelming. In real life, one light-water reactor plant produces about 1GW of power. On average, these plants have 4 reactors each, meaning the raw output of each reactor is something like ~250MW. In Factorio, one plant of 4 reactors produces under 500MW, counting the magic adjacency bonus with maximum steam turbines. So, to get the same power output, twice the amount of reactors are needed. Using the Factorio Cheat Sheet as a reference reveals that adding more reactors just gives diminishing returns to the power output ratio as well.

In addition to this, nuclear power setups use tons of water. Taking water in by train is less than ideal in most cases since so much water is consumed. In fact, some friends and I were jokingly talking about units of consumed water in wagons/second. If you can't take the water in by train, you need to find water nearby to run pipes to your heat exchangers. Having enough offshore pumps can be difficult in many cases where heat exchangers have to be tightly packed to make the most efficient use of the heat coming off the reactor. Also, any pipes between heat exchangers and/or turbines means that you're limiting the throughput of steam to your turbines due to the current state of the fluid mechanics. 1 heat exchanger to 1.67 turbines is not a very clean ratio for maximizing efficiency, either.

The reactors themselves do not lend themselves any extra utility besides generating heat. There is no way to directly control the reactor's state using the circuit network. You can use the circuit network right up until the final inserter where the fuel cell is put in, but without a way to read contents, the best way to ration your fuel is to either count how many fuel cells you're putting in or time it accordingly. Too many fuel cells means eating through more of that hard-earned U-235 if consumption is not properly limited. Now, I don't have a problem with the warm-up and cool-down times of the reactor since it adds an extra layer of challenge to making a waste-free reactor, but not having direct control of the reactor's consumption through the circuit network just feels wrong when basically every other production building can jack in.

Nuclear power takes a lot of setup for such an underwhelming payoff. This extra setup is simply something that people tend to avoid unless they really want to do nuclear power. I personally love nuclear power, but I have to set it up myself all the time when playing with friends because it's such a hassle that nobody else wants to bother with. I don't really have a problem with the extra time and effort that goes into getting nuclear mining and processing done since I personally find it fun. The problem is really just how expensive it is when you hardly benefit from it, especially with the other players mindlessly spamming solar energy. Nuclear needs a huge buff to stay relevant in the current meta.

Alright, nuclear power needs a buff. How can it be improved?
The way I see it, we have several options:

• Flat increase to power generation from turbines (multiply output by a large amount, possibly x4 or x8)
• Better effective ratio of heat exchangers to turbines (the same as regular Steam, 1-2 vs. the current 1-1.67)
• Better ratios of pumps to heat exchangers with by greatly reducing water consumption (1-24 or 1-48 vs. the current 1-12)
• Overall optimizations to UPS to make nuclear power as a whole more viable in the endgame
• Better integration of nuclear power structures into the circuit network
• ^ All of the above!

Conclusion
I really, really love nuclear power. I don't see any legitimate reason to keep it the way it is in its current state, especially since nobody is screaming that it's OP, and I don't think anyone would throw a fit if it was buffed a bit more. I think in a game like Factorio, nuclear power just fits very well. Although it's been marvelously implemented, it just needs some tweaks to get it in a better state of affairs.

This is my love letter to nuclear power in Factorio. Glorious developers, hear my rallying cry and bring out the best of your greatest content addition since trains. Make nuclear power great again!

[Koub]

Introduction
So, we hear it all the time: "Solar power is waaaay better than nuclear."

I have heard it some times, but saying it's heard all the time is a blatant exageration.

Solar produces massive amounts of energy per tile to the point where you have to suspend your disbelief to use it.

There isn't a single element of the game that doesn't need some disbelief suspension. Fortunately, the only few aspects that request significantly more disbelief suspension than the rest of the game do so for gameplay purpose (magic water pump, magic belts, magic inventory, ...).

Also, I'm convinced that UPS are the worst reason to buff (or nerf) something. I'm not saying nuclear should not be buffed. I'm not saying it should be either, but your arguments don't convince me the game needs a nuclear buff.

[Tesse11ation]

I have heard it some times, but saying it's heard all the time is a blatant exageration.

Is it really that blatant though? Sure it's a hyperbole, but if enough people believe it, it's probably true. I kind of needed a declarative statement to open with as well, from a writing structure standpoint. Most declarative statements are a hyperbole anyway.

There isn't a single element of the game that doesn't need some disbelief suspension. Fortunately, the only few aspects that request significantly more disbelief suspension than the rest of the game do so for gameplay purpose (magic water pump, magic belts, magic inventory, ...).

I admit, suspension of disbelief is subjective. I mostly said that as a S/O to the guy who actually took a minute to do the math just for fun. I could extrapolate on some of your other points, but as nuclear is the focus, I'd rather not do that here. Furthermore, in the proceeding sentence after this I declare this to be an acceptable break from reality since Factorio is a video game and its primary goal is fun for the player.

Also, I'm convinced that UPS are the worst reason to buff (or nerf) something. I'm not saying nuclear should not be buffed. I'm not saying it should be either, but your arguments don't convince me the game needs a nuclear buff.

From a gameplay standpoint, I think you're absolutely right. I just find it interesting that of all the points I brought up, none of the others were convincing enough, and this looks to be the dealbreaker for you. From a practical standpoint, like it or not, UPS is god once you hit endgame and go full megabase. Even without UPS taken into consideration there's still quite a bit of balance to be done IMO. Not to mention fewer buildings for the same output would be a de facto fix for UPS consumption, which is more the direction I'm leaning towards for this leg of the argument anyway.

[eradicator]

So, we hear it all the time: "Solar power is waaaay better than nuclear."

Yea well, when they say "better" they mean "faster" though. As you said yourself, UPS is king.

Late-game, solar has a negligible UPS cost for its size.

No, it's not "negligible", it's O(1), it's a fixed cost. UPS cost of solars is independant of the number of cells you have. THIS is why megafactories use it, and this is why nothing can beat for megafactories. No amount of optimization or buffing can make nuclear cost independant of the number of reactors. Even 1 reactor, 1 heat exchanger and 1 turbine are more cpu-expensive than 1TW of solar. (EDIT: see FFF 148)

But...nuclear as it is now is a really good solution for non-megafactories (which is 99% of all players), getting you a few GW of power with way less effort than solar needs for that.

[mrvn]

What makes solar better?
What else makes solar energy so powerful? Updates per second. Late-game, solar has a negligible UPS cost for its size. This is great, but it also means that by late-game, nuclear simply is not viable by comparison. This has to be by far the number one reason why nobody... and I mean NOBODY... uses nuclear for their megabase.

Solar power doesn't have a negligible UPS cost. It has a constant UPS cost no matter the size. The game counts up all solar cells connected to an electrical network ONCE and then each tick you get <power per cell> * <number of cells> of power. Could even be that even if you have no solar cells you still get <power per cell> * 0 amount of power every tick. Meaning even with 0 cells the same code gets executed, just with a multiplier of 0. The extra check weather you have solar cells or not would cost more time, solar cells are that cheap for UPS.

Why does nuclear power suck?
Aside from the UPS issues, nuclear power is actually quite underwhelming. In real life, one light-water reactor plant produces about 1GW of power. On average, these plants have 4 reactors each, meaning the raw output of each reactor is something like ~250MW. In Factorio, one plant of 4 reactors produces under 500MW, counting the magic adjacency bonus with maximum steam turbines. So, to get the same power output, twice the amount of reactors are needed. Using the Factorio Cheat Sheet as a reference reveals that adding more reactors just gives diminishing returns to the power output ratio as well.

How is there a diminishing return? The optimal setup is building reactors in a 2*N grid and besides the 4 reactors on the sides every single reactor in the middle gives the same amount of energy. Every pair you add adds the exact same amount. It adds a constant amount, not a diminishing amount. It's the 4 reactors at the sides that are suboptimal since they have less neighbor bonus. So the larger N is the closer you get to the optimal output.

In addition to this, nuclear power setups use tons of water. Taking water in by train is less than ideal in most cases since so much water is consumed. In fact, some friends and I were jokingly talking about units of consumed water in wagons/second. If you can't take the water in by train, you need to find water nearby to run pipes to your heat exchangers. Having enough offshore pumps can be difficult in many cases where heat exchangers have to be tightly packed to make the most efficient use of the heat coming off the reactor. Also, any pipes between heat exchangers and/or turbines means that you're limiting the throughput of steam to your turbines due to the current state of the fluid mechanics. 1 heat exchanger to 1.67 turbines is not a very clean ratio for maximizing efficiency, either.

The reactors themselves do not lend themselves any extra utility besides generating heat. There is no way to directly control the reactor's state using the circuit network. You can use the circuit network right up until the final inserter where the fuel cell is put in, but without a way to read contents, the best way to ration your fuel is to either count how many fuel cells you're putting in or time it accordingly. Too many fuel cells means eating through more of that hard-earned U-235 if consumption is not properly limited. Now, I don't have a problem with the warm-up and cool-down times of the reactor since it adds an extra layer of challenge to making a waste-free reactor, but not having direct control of the reactor's consumption through the circuit network just feels wrong when basically every other production building can jack in.

Nuclear power takes a lot of setup for such an underwhelming payoff. This extra setup is simply something that people tend to avoid unless they really want to do nuclear power. I personally love nuclear power, but I have to set it up myself all the time when playing with friends because it's such a hassle that nobody else wants to bother with. I don't really have a problem with the extra time and effort that goes into getting nuclear mining and processing done since I personally find it fun. The problem is really just how expensive it is when you hardly benefit from it, especially with the other players mindlessly spamming solar energy. Nuclear needs a huge buff to stay relevant in the current meta.

Alright, nuclear power needs a buff. How can it be improved?
The way I see it, we have several options:

• Flat increase to power generation from turbines (multiply output by a large amount, possibly x4 or x8)
• Better effective ratio of heat exchangers to turbines (the same as regular Steam, 1-2 vs. the current 1-1.67)
• Better ratios of pumps to heat exchangers with by greatly reducing water consumption (1-24 or 1-48 vs. the current 1-12)
• Overall optimizations to UPS to make nuclear power as a whole more viable in the endgame
• Better integration of nuclear power structures into the circuit network
• ^ All of the above!

Conclusion
I really, really love nuclear power. I don't see any legitimate reason to keep it the way it is in its current state, especially since nobody is screaming that it's OP, and I don't think anyone would throw a fit if it was buffed a bit more. I think in a game like Factorio, nuclear power just fits very well. Although it's been marvelously implemented, it just needs some tweaks to get it in a better state of affairs.

This is my love letter to nuclear power in Factorio. Glorious developers, hear my rallying cry and bring out the best of your greatest content addition since trains. Make nuclear power great again!

One way nuclear power gets a buff with mods is to give it extra uses. For example the reactor is used to enrich uranium / make plutonium used in atom bombs and nuclear fuel. So to make big BOOMs or have fast trains you have to have running reactors. That means you at least need the reactors even if you skip the heat exchangers and steam turbines. On the other hand this makes the circuit logic irrelevant. At least so far I haven't reached the point where I could slow down the reactors because my nuclear fuel backs up.

Personally I think in vanilla the water and steam throughput is the biggest problem. In a large 2*N array every reactor produces more steam than a pipe without pumps can transport and needs 3 water pumps. Maybe something other than steam could be used to reduce the amount per W or a semi closed system where only some of the water is lost and some is recovered.

[Ultros]

I totally use nuclear for my megabase. It's not a 100% efficient megabase, but it still does 2k spm with biters at a respectable UPS.

This is a game, and it's intended to be fun. For me, nuclear is way more fun than placing down a million solar panels and accumulators. That's reason enough to use it. The fun is in the setup and automation of a well built nuclear power setup.

I feel nuclear is pretty well balanced as-is. Regarding the water issue, this is pretty realistic as the best way to build NPPs ingame is to build them next to or on top of a lake, as they are in real life. It makes perfect sense that they consume ungodly amounts of water, it's not realistic to train in water enough for multiple GW of nuclear power.

The "power wasted" issue can be resolved by designing a "smart reactor" which uses steam storage and circuitry to ensure that fuel is only inserted when it is required. These blueprints are common, although not UPS efficient, and the best UPS efficient reactors are used in megabases anyway, where the power consumption is pretty constant.

[mrvn]

I totally use nuclear for my megabase. It's not a 100% efficient megabase, but it still does 2k spm with biters at a respectable UPS.

This is a game, and it's intended to be fun. For me, nuclear is way more fun than placing down a million solar panels and accumulators. That's reason enough to use it. The fun is in the setup and automation of a well built nuclear power setup.

I feel nuclear is pretty well balanced as-is. Regarding the water issue, this is pretty realistic as the best way to build NPPs ingame is to build them next to or on top of a lake, as they are in real life. It makes perfect sense that they consume ungodly amounts of water, it's not realistic to train in water enough for multiple GW of nuclear power.

The "power wasted" issue can be resolved by designing a "smart reactor" which uses steam storage and circuitry to ensure that fuel is only inserted when it is required. These blueprints are common, although not UPS efficient, and the best UPS efficient reactors are used in megabases anyway, where the power consumption is pretty constant.

A smart reactor needs one steam tank connected to one of the middle lines of heat exchangers and steam turbines. You don't even need a steam tank per reactor. The effect on UPS for that one added tank should be negible.

[Ranger_Aurelien]

A smart reactor needs one steam tank connected to one of the middle lines of heat exchangers and steam turbines. You don't even need a steam tank per reactor. The effect on UPS for that one added tank should be negible.

A nuclear power "facility" is to generate consistent electricity to the base. In factorio the front line is turbines "burning" steam to produce electricity -- based on electrical load! If there is no demand, no steam is consumed*. In that case, there is no need to heat the nuclear core.


Also note a balancing factor for solar power (+accumulators) is surface area. If playing defaults-set+vanilla Factorio(tm), surface area comes at a "cost" of destroying biter bases which advances evolution greatly. In post-rocket worlds I've built the solar+accum footprint is several times larger than the factory part of the base (not accounting for defensive components). (I suppose an up-side is the solar area can act as a buffer to stop aggro-ing biters**.)

* Steam is a fluid; you could even make it at a satellite plant, pipe it to trains, stockpile it in your base for when it is needed!

** (Yes, there are mods to make your Pollution's cloud diminish solar under it...).

[slippycheeze]

A smart reactor needs one steam tank connected to one of the middle lines of heat exchangers and steam turbines. You don't even need a steam tank per reactor. The effect on UPS for that one added tank should be negible.

A nuclear power "facility" is to generate consistent electricity to the base. In factorio the front line is turbines "burning" steam to produce electricity -- based on electrical load! If there is no demand, no steam is consumed*. In that case, there is no need to heat the nuclear core.

I agree with this: nuclear in Factorio generates power for a fixed amount of time when fuel is inserted. You should use it to fill an appropriately sized steam storage facility (which, incidentally, has a vastly higher energy capacity per tile than accumulators) which is then used to drive turbines for power generation.

Insert nuclear fuel to fill the storage as needed, "withdraw" to power turbines as needed, and don't try and directly power anything from nuclear. If it generates fixed amounts of output, which it does, then you want to capture that and emit it at the needed rate, smoothing the bursty generation into a nice, slow consumption.

[someone1337]

No, it's not "negligible", it's O(1), it's a fixed cost. UPS cost of solars is independant of the number of cells you have. THIS is why megafactories use it, and this is why nothing can beat for megafactories. No amount of optimization or buffing can make nuclear cost independant of the number of reactors. Even 1 reactor, 1 heat exchanger and 1 turbine are more cpu-expensive than 1TW of solar.

Is solar REALLY O(1)?

The solar panels may be, but are the accumulators also O(1)?

Also, if power becomes THE major UPS bottleneck, you can always go for an electric energy interface

* Steam is a fluid; you could even make it at a satellite plant, pipe it to trains, stockpile it in your base for when it is needed!

It is imho always cheaper to run one fucking large electricity network all over the map and have just one central power location.

[slippycheeze]

No, it's not "negligible", it's O(1), it's a fixed cost. UPS cost of solars is independant of the number of cells you have. THIS is why megafactories use it, and this is why nothing can beat for megafactories. No amount of optimization or buffing can make nuclear cost independant of the number of reactors. Even 1 reactor, 1 heat exchanger and 1 turbine are more cpu-expensive than 1TW of solar.

Is solar REALLY O(1)?

Yes, as O(1) really means O(1*N) where N is the number of entities. In this case, separate power networks with solar connected.

The solar panels may be, but are the accumulators also O(1)?

Yes, as above.

[slippycheeze]

* Steam is a fluid; you could even make it at a satellite plant, pipe it to trains, stockpile it in your base for when it is needed!

It is imho always cheaper to run one fucking large electricity network all over the map and have just one central power location.

I'm not saying it is a good idea, just a possible one. Distributed generation has advantages as well as costs, though: one network, one chompi boi, one entire base out of power, and now you got many chompi boi to deal with.

[Ultros]

Yes, as O(1) really means O(1*N) where N is the number of entities. In this case, separate power networks with solar connected.

This makes no sense given the paradigm of big O notation. eradicator's previous is 100% correct. "O(1*N)" is defined as O(N), not O(1) as you are saying.

Solar panels and accumulators are O(1) when applied to UPS, because the capacity of the network of connected panels/accumulators is calculated when new entities are added/removed from the network, and this value is cached and used for power production and draw, consuming a constant amount of computation every tick. The number of entities doesn't change the compute time required per frame, therefore the UPS effect is O(1).

[Koub]

[Koub] Please stay on topic. You can take for granted solar + accus are free UPS-wise within a given solar network, no matter how many solar panels and accus.

[mrvn]

A smart reactor needs one steam tank connected to one of the middle lines of heat exchangers and steam turbines. You don't even need a steam tank per reactor. The effect on UPS for that one added tank should be negible.

A nuclear power "facility" is to generate consistent electricity to the base. In factorio the front line is turbines "burning" steam to produce electricity -- based on electrical load! If there is no demand, no steam is consumed*. In that case, there is no need to heat the nuclear core.

I agree with this: nuclear in Factorio generates power for a fixed amount of time when fuel is inserted. You should use it to fill an appropriately sized steam storage facility (which, incidentally, has a vastly higher energy capacity per tile than accumulators) which is then used to drive turbines for power generation.

Insert nuclear fuel to fill the storage as needed, "withdraw" to power turbines as needed, and don't try and directly power anything from nuclear. If it generates fixed amounts of output, which it does, then you want to capture that and emit it at the needed rate, smoothing the bursty generation into a nice, slow consumption.

Heat pipes are even more efficient than steam tanks at storing energy. And when it comes to UPS you don't want a big steam storage setup.

As said you only need one steam tank. You can't measure the temperature of the reactor to see when it needs a new fuel cell so you have to measure the steam production to deduce the temperature. As long as the reactor is hot enough to produce more steam than you consume the steam level will not fall. Once it falls it's time to throw some fuel on the fire.

[coppercoil]

In addition to this, nuclear power setups use tons of water. Taking water in by train is less than ideal in most cases since so much water is consumed. In fact, some friends and I were jokingly talking about units of consumed water in wagons/second.

I wouldn't say it's a joke. If you’re building a megabase, then tons of smth per second is not funny, it’s your challenge. High throughput water delivery rail network must be totally different. 20L40C trains… circuit-controlled junctions with multiple waiting lanes… It sounds cool! Maybe I’ll try it . I love nuclear power.
Nevertheless, I agree that water delivery using pipes should not be an issue for players who are looking for another kind of challenges.

[Tesse11ation]

Ok, I get it with what people have to say about UPS. It's my fault for putting that argument foremost instead of the water consumption statistics, as it's much less pertinent IMO. So, I'd like to add some hard numbers since I pretty much neglected to do this before.

This next calculation just works to reinforce my point -- the amount of water required for nuclear is not just absurd, but also makes no sense from a realistic point of view. Let's look at fuel value vs. output. Doing the math for the power output of regular steam engines vs. nuclear turbines at maximum capacity reveals a single major discrepancy:

Coal-fueled boiler: 1 MJ = ~33.33 units of water, /165 degrees C = ~200 kJ per temperature unit of Steam
Uranium fuel cell: 1 MJ = ~1 unit of water, /500 degrees C = ~2 kJ per temperature unit of Steam

I'll let the numbers speak for themselves. Knowing this, we can be fairly certain that dividing water consumption of heat exchangers by 100 will give us consistent results when compared to the regular steam alternative, unless my math is incorrect or I missed a detail somewhere. You can't tell me based on this that coal power is magically more efficient than nuclear in any scenario where the laws of physics as we know them apply.

[someone1337]

No, it's not "negligible", it's O(1), it's a fixed cost. UPS cost of solars is independant of the number of cells you have. THIS is why megafactories use it, and this is why nothing can beat for megafactories. No amount of optimization or buffing can make nuclear cost independant of the number of reactors. Even 1 reactor, 1 heat exchanger and 1 turbine are more cpu-expensive than 1TW of solar.

Is solar REALLY O(1)?

Yes, as O(1) really means O(1*N) where N is the number of entities. In this case, separate power networks with solar connected.

The solar panels may be, but are the accumulators also O(1)?

Yes, as above.

You are describing O(n) - theta notation is about describing the worst case

A freshly built solar panel gives the current day's state of power, a freshly built accumulator charges diffrently at first and lets assume you never fully charge and fully discharge them, i would expect that you cannot assume they can be regarded as a single entity (unlike solar panels)

Anyway: ontopic... Imho the biggest issue with nuclear is its ups hungryness.
I tried to get that a little lower by writing a mod, which would have created one 100*100-sized entity, which was basically a screenshot of my nuclear powerplant and wanted it ro take nuclear cells and water and produce power.
The project is currently on hold, because it was not possible with 0.16 to create a single tile that big

Followed by: there is far too much uranium on the map, so it has to be inefficient as hell, as you woukd just be able to run with one cell for waaaay too long.

Maybe the 1 uranium -> 10 cells could get changed a little to like 1:1, but make the reactors more efficient.

[eradicator]

i would expect that you cannot assume they can be regarded as a single entity (unlike solar panels)

It was stated that as as soon as a freshly built accumulator has fully charged (or discharged) and is thus in sync with "the network" it will be joined into the "singular group" of optimized accumulators for that network. You are right that accus have to reach synchronization first, but that happens at least once a day (~7 real-life minutes) in a normal factory. If you have a basic understanding of the modding api you can use this command to see that accumulators have an otherwise unused priority. If ofc you build your factory on purpose to never fully dis/charge any accumulators you might be able to "evade" getting them optimized.

Code: Select all

/c game.player.print(game.player.selected.prototype.electric_energy_source_prototype.usage_priority)

(Btw, i edited the relevant solar FFF into above post because Koub said discussing solar was off topic :p.)

____
Ok, let's throw around some more facts and formulas:

• There's only one type of "steam":

  Code: Select all

    {
      type = "fluid",
      name = "steam",
      default_temperature = 15,
      max_temperature = 1000,
      heat_capacity = "0.2KJ",
      gas_temperature = 15,
    },
  

• Both coal-boiler and reactor are defined as having fuel burning efficiency of 100%.
• Reactors get additional "free" energy from adjacency.
• (500-15)/(165-15) * (2*900kW) == 5.82MW, i.e. a "Steam engine" has the same efficiency as a "Steam turbine".
• (500-15) * 0.2kJ * 60 == 5820MJ, production of turbine
• (165-15) * 0.2kJ * 30 == 900MJ, production of engine

So, i'm not sure at what point you're saying nuclear supposedly loses efficiency. Btw, you can run a steam turbine on a coal boiler and get the expected 1.8MW of electricity.

[Tesse11ation]

Ok, let's throw around some more facts and formulas:

• There's only one type of "steam":

  Code: Select all

    {
      type = "fluid",
      name = "steam",
      default_temperature = 15,
      max_temperature = 1000,
      heat_capacity = "0.2KJ",
      gas_temperature = 15,
    },
  

• Both coal-boiler and reactor are defined as having fuel burning efficiency of 100%.
• Reactors get additional "free" energy from adjacency.
• (500-15)/(165-15) * (2*900kW) == 5.82MW, i.e. a "Steam engine" has the same efficiency as a "Steam turbine".
• (500-15) * 0.2kJ * 60 == 5820MJ, production of turbine
• (165-15) * 0.2kJ * 30 == 900MJ, production of engine

So, i'm not sure at what point you're saying nuclear supposedly loses efficiency. Btw, you can run a steam turbine on a coal boiler and get the expected 1.8MW of electricity.

Alright, I see my math is incorrect. I didn't have time to proofread when I posted. I think actually that the discrepancy (if there is one) is the MJ value of the fuel type against total water consumed, and the case to be made is that the water cost of nuclear is just way too damned much compared to regular steam.

I would break it down more if I had the time to do so. My main point here is that the one thing that nuclear eats through faster than UPS is water.