Nuclear power ratios

The four numbers behind everything

Entity	Statistic	Value
Nuclear reactor	Heat output	40 MW
Nuclear reactor	Bonus per adjacent reactor	+100%
Heat exchanger	Heat consumed	10 MW
Steam turbine	Steam consumed	60.0 / s

Where the awkward number comes from

A heat exchanger turns water at 15 degrees into steam at 500 degrees. Steam holds 200 joules per unit per degree, so raising one unit across that 485 degree gap takes 97 kJ. An exchanger pushing 10 MW therefore produces 103.1 steam per second.

A turbine consumes 60.0 steam per second. Divide, and one exchanger supports 1.718 turbines. There is the awkward number: it is 10 MW divided by the 5.82 MW a turbine can actually convert. The figure stays awkward because 485 shares no convenient factor with the turbine's consumption rate.

Scaling that to one unbonused reactor gives the familiar starting block: 4 exchangers and 6.873 turbines. The advice to build seven is a rounding decision, not the ratio.

Reactor blocks and the neighbour bonus

Each reactor produces its base output again for every adjacent reactor it touches. In a 2 by N row every reactor touches at least two others, and the interior ones touch three.

Block	Reactors	Heat output	Multiplier	Exchangers	Turbines
2 x 1	2	160 MW	2.00x	16	27.49
2 x 2	4	480 MW	3.00x	48	82.47
2 x 3	6	800 MW	3.33x	80	137.46
2 x 4	8	1120 MW	3.50x	112	192.44
2 x 6	12	1760 MW	3.67x	176	302.41
2 x 8	16	2400 MW	3.75x	240	412.37

Why the 2 by N shortcut works

Every one of those rows satisfies a simple closed form: total output in megawatts equals 160n − 160, where n is the reactor count. The reason is visible in the adjacency count. Adding a pair of reactors to the end of a 2 by N row creates four new touching pairs, and each pair is worth 40 MW to each of its two reactors — so every added pair contributes 80 MW of base output plus 320 MW of bonus, a flat 160 MW per reactor added. The constant −160 is the bonus the first pair does not receive, having nothing to its left.

This is also why long rows beat compact squares. A 2 by 2 block averages 3.00x per reactor; a 2 by 8 averages 3.75x. The multiplier climbs towards 4 without reaching it: in any finite 2 by N row the four reactors at the two ends touch one fewer neighbour than the interior ones, so some bonus is left unclaimed at every length.

Rounding: when it costs fuel and when it does not

A reactor burns its fuel cell at a fixed rate regardless of demand, so the cost of getting the ratio wrong is not symmetric:

• Too few turbines and steam backs up. Exchangers stop drawing heat, the reactor hits its temperature ceiling, and the excess is discarded — you burn a full cell and throw part of it away.
• Too many turbines and the surplus ones simply sit idle most of the time. They cost materials and space, not fuel.

Given that asymmetry, rounding the turbine count up is the cheaper mistake. Rounding the exchanger count up does nothing at all: exchangers are limited by available heat, so extra ones cannot draw more than the reactors produce.

Why 2 by N beats the square it appears to lose to

The neighbour bonus math seems to favour compact squares. A 4 by 4 block averages 4.00x per reactor; a 2 by 8 row of the same 16 reactors averages 3.75x. On paper the square produces 160 MW more. The catch is that not all of that output is reachable.

Layout	Reactors	Theoretical output	Avg multiplier	Landlocked reactors	Edge reactors
2 x 2	4	480 MW	3.00x	0	4
3 x 3	9	1320 MW	3.67x	1	8
2 x 4	8	1120 MW	3.50x	0	8
4 x 4	16	2560 MW	4.00x	4	12
2 x 8	16	2400 MW	3.75x	0	16
2 x 6	12	1760 MW	3.67x	0	12

A reactor is landlocked when all four of its sides touch another reactor. In a 4 by 4 block, the four interior reactors have no exposed face. Heat exchangers cannot sit adjacent to them, so their heat must travel out through heat pipes before it can be converted to steam. Heat pipes lose temperature over distance, and the prototype data in data/2.1/ does not include heat pipe attenuation values. That means the exact distance at which a landlocked reactor's heat fails to reach an exchanger above working temperature cannot be computed from the available data.

A 2 by N row has zero landlocked reactors at any length. Every reactor has at least one long side exposed, and heat exchangers can be placed directly against that side. The heat travels zero tiles through a pipe. The trade-off is that some neighbour bonus is left unclaimed compared to a square, but the output that remains is all extractable without betting on heat pipe behaviour the data does not specify.

This is why the closed form 160n − 160 is stated for 2 by N rows specifically. It is not the maximum theoretical output for n reactors; it is the maximum output that every reactor can deliver directly to an adjacent exchanger, with no heat pipe in the critical path. A square can produce more on paper, but realising that surplus requires solving a heat distribution problem whose parameters are not in the data set this page is built from.

The fuel cycle and what Kovarex actually changes

Reactor ratios describe how to turn heat into electricity. They say nothing about where the uranium comes from or how long it lasts. The fuel side has its own chain, and the numbers that govern it are all in recipes.json.

Step	Input	Output	Time
Uranium processing	10 uranium ore	0.7% U-235 / 99.3% U-238 (one isotope)	12 s
Fuel cell assembly	1 U-235 + 19 U-238 + 10 iron plate	10 uranium fuel cells	10 s
Reprocessing	5 depleted fuel cells	3 U-238	60 s
Kovarex enrichment	40 U-235 + 5 U-238	41 U-235 + 2 U-238	60 s

Uranium processing: the isotope bottleneck

Every centrifuge cycle consumes 10 uranium ore and produces one isotope: U-235 with 0.7% probability, otherwise U-238. That means roughly 143 cycles, or about 1,429 ore, are needed to produce a single U-235 on average. U-238 is produced at roughly 142 times the rate of U-235 from raw ore.

A fuel cell requires 1 U-235 and 19 U-238 and yields 10 cells. The 19-to-1 ratio is where the imbalance bites: ore processing produces U-238 at about 142 times the U-235 rate, so U-238 accumulates rapidly while U-235 remains the constraint. Without reprocessing or enrichment, every 10 cells permanently consumes 1 U-235 and 19 U-238, and the U-238 stock only grows.

Reprocessing closes the U-238 loop

Spent fuel cells are not waste. Five depleted cells reprocess into 3 U-238 over 60 seconds in a centrifuge, returning 0.6 U-238 per cell. That reduces the net U-238 cost per fuel cell from 1.9 to 1.3. It does not return any U-235. Reprocessing is therefore a U-238 recycling step, not a way to stretch fissile material.

At a steady consumption rate of one fuel cell per second, one centrifuge can assemble cells fast enough (10 cells per 10-second craft), but reprocessing the spent cells requires 12 centrifuges because each reprocessing cycle handles only 5 cells in 60 seconds. The reprocessing chain is therefore much larger than the assembly chain for the same throughput.

Kovarex enrichment turns the surplus into fuel

The Kovarex process consumes 40 U-235 and 5 U-238 and produces 41 U-235 and 2 U-238 over 60 seconds. The 40 U-235 input is a catalyst stock, not a consumption: it is returned in full, and the process adds one net U-235 while consuming three U-238. It requires a bootstrap of 40 U-235 to start, after which it runs indefinitely as long as U-238 is supplied.

This is the step that makes nuclear self-sustaining. Without it, U-235 production is throttled by the 0.7% ore processing probability, and U-238 piles up with no outlet. With Kovarex running, the U-238 surplus is converted into additional U-235 at a rate of 3 U-238 per 1 U-235 per 60-second cycle. The reactor's neighbour bonus increases heat output per cell but does not change the rate at which cells are consumed, so the fuel chain scales with reactor count, not with block layout.

One important data limitation: the fuel value of a uranium fuel cell (how long one cell burns in a reactor) is not present in items.json or machines.json. The reactor's burner source accepts the nuclear fuel category, but the item record for uranium-fuel-cell carries no fuel_value field. This means the exact burn duration and therefore the precise reactor-to-centrifuge ratio cannot be computed from the available data. The recipe-side ratios above are complete; the consumption rate is not.

Heat pipe attenuation: the dimension the data cannot answer

Heat pipes carry reactor heat to exchangers. In a long 2 by N row with exchangers placed directly against reactor faces, the critical heat path is zero or one pipe long and attenuation is not a concern. In any layout where heat must travel through multiple pipe segments to reach an exchanger, temperature drops along the way. If the temperature at an exchanger falls below 500 °C, it stops producing steam even though the reactor block is generating ample total heat.

The prototype data available for this page contains no heat pipe entity at all. There is no entry for heat pipe thermal capacity, heat transfer rate, or per-tick temperature loss in machines.json, items.json, or any other file in data/2.1/. The heat exchanger record does include a maximum heat transfer of 2 GW and a minimum working temperature of 500 °C, but the pipe that connects reactor to exchanger has no corresponding prototype data in this data set.

The practical implication is straightforward: keep heat pipes short. Place exchangers directly against reactor faces whenever possible, and if a layout requires longer runs, build it and observe exchanger temperatures rather than trusting a calculation the data cannot support. This page can give exact ratios for reactors, exchangers and turbines because all four prototype values are present. It cannot give a maximum heat pipe distance because that value is not in the data.

Concrete rounding for each block size

The asymmetry between too few and too many turbines is stated above. Here is what it means at each standard block size, using the exact turbine counts from the table:

• 1 reactor (unbonused): 6.87 turbines. Build 7. The 0.13-turbine surplus is idle capacity; building 6 would back up steam and waste fuel.
• 2 by 1 (2 reactors, 160 MW): 27.49 turbines. Build 28.
• 2 by 4 (8 reactors, 1,120 MW): 192.44 turbines. Build 193.
• 2 by 8 (16 reactors, 2,400 MW): 412.37 turbines. Build 413.

The fractional surplus does not shrink monotonically — it depends on where the 1.718 multiplier lands — but the rule does not change. Rounding up costs a small amount of iron and space for the idle turbine, while rounding down wastes uranium fuel cells for as long as the plant runs. Over the life of a megabase, the fuel cost dwarfs the turbine cost, which is why the rule is always round up.

Exchangers do not have this asymmetry. A single reactor produces exactly 4 exchangers' worth of heat (40 MW / 10 MW), but for larger blocks the exchanger count is always a whole number because the neighbour bonus produces heat in multiples of 40 MW. Exchangers are limited by available heat: an extra exchanger simply does not activate if no surplus heat exists. Building 5 instead of 4.12 wastes the materials for one exchanger but has no runtime cost, unlike a missing turbine.

What the missing fuel value means for planning

The exact burn time of one fuel cell is not in the available prototype data, so this page cannot tell you how many fuel cells a reactor consumes per hour or how many centrifuges you need to sustain a specific block. What can be stated from the data is the recipe-side economy: each cell requires 0.1 U-235 and 1.9 U-238 to assemble (1.3 net after reprocessing), and Kovarex converts 3 U-238 into 1 net U-235 per 60-second cycle.

In practice, players who need an exact reactor-to-centrifuge ratio should measure the burn rate in-game by placing a fixed number of cells in a reactor and timing how long they last, then dividing the reactor count into the cell consumption rate. That measurement, combined with the recipe ratios above, gives the complete fuel chain sizing. The heat-to-electricity ratios on this page are exact; the fuel burn rate requires one empirical measurement that the prototype data does not provide.

Applies when…

• The neighbour bonus counts only reactors that are actively burning fuel and directly adjacent. A partly fuelled block produces less than these figures, and an unfuelled neighbour contributes nothing.
• Turbine counts assume steam actually reaches every turbine. Pipe throughput limits apply separately and are the usual reason a correctly-sized plant underperforms.
• Figures are for normal quality equipment. Quality affects several of these statistics.
• Heat must reach the exchangers above the working temperature. Long heat pipe runs lose temperature and can leave distant exchangers idle even when total heat output is sufficient.
• These are uranium fission figures. The fusion system introduced on Aquilo is a separate mechanism and is not covered here.

Related

• Solar panels and accumulators — the other way to power a base
• Production Ratio Calculator — how much power a line actually draws
• All ratio pages