Smelting ratios

Furnaces per belt of ore

A plate takes 3.2 seconds to smelt, divided by the furnace's crafting speed. A belt delivers a fixed number of ore per second. Divide one by the other and you have the column length.

Belt	Ore / s	stone furnace	steel furnace	electric furnace
transport belt	15.0	48.0	24.0	24.0
fast transport belt	30.0	96.0	48.0	48.0
express transport belt	45.0	144.0	72.0	72.0
turbo transport belt	60.0	192.0	96.0	96.0

The familiar 24-furnace column comes straight out of this: a transport belt carries 15.0 ore per second, and a furnace at speed 2 clears one every 1.6 seconds. Twelve per side of the belt.

Which furnace tier, and why it is not about speed

Steel and electric furnaces run at the same speed, so the tier decision comes down to two other things: energy per plate, and module slots.

Furnace	Speed	Power draw	Plates / s each	Energy per plate	Module slots
stone furnace	1	90 kW	0.313	288.0 kJ	0
steel furnace	2	90 kW	0.625	144.0 kJ	0
electric furnace	2	180 kW	0.625	288.0 kJ	2

On energy per plate the steel furnace wins at 144.0 kJ. The electric furnace costs 2.0 times as much energy per plate for identical output — it earns its place through the 2 module slots, not through throughput.

That reframes the upgrade decision. Moving from steel to electric furnaces without fitting modules is a pure loss: same plates, more power, more materials. The upgrade only pays once productivity or speed modules go into those slots.

The steel ratio that makes columns easy

Steel takes 16 seconds per craft and consumes 5 iron plates. Iron takes 3.2 seconds. Those numbers are not independent: 16 divided by 3.2 is exactly 5, the same as the ingredient count.

The consequence is a ratio worth remembering because it needs no arithmetic: one steel furnace per iron furnace, at any furnace tier, as long as both tiers match. A steel furnace consumes iron plates at precisely the rate one iron furnace produces them.

Step	Machine	Count per 1 steel/s
steel plate	electric furnace	8.00
iron plate	electric furnace	8.00

The counts are equal, as the ratio predicts. A steel column is therefore a mirror of the iron column feeding it, which is why steel builds tile so cleanly next to iron ones.

What the two module slots are worth

An unmodified electric furnace draws 180 kW for the same output as a steel furnace at 90 kW. The electric furnace costs twice the energy for identical throughput, so without modules the upgrade is a pure loss on power. The justification for the upgrade is entirely in those 2 module slots. The question is: what do you have to put in them before the electric furnace earns its place?

Each module modifies three things at once: crafting speed, energy draw, and (for productivity modules) free extra output. The numbers below use the effects straight from the item prototypes. Speed module effects stack additively: two speed module 3s add +100% speed, not +125%. Productivity modules add a speed penalty that also stacks. Energy usage cannot drop below 20% of the machine's base draw, which is why two efficiency module 3s floor at 36 kW rather than reaching zero.

Module configuration	Effective speed	Power draw	Plates / s	Energy / plate	Ore / plate	Furnaces per yellow belt
No modules	2.00	180 kW	0.6250	288.0 kJ	1.000	24.0
2 × speed module	2.80	360 kW	0.8750	411.4 kJ	1.000	17.1
2 × speed module 2	3.20	396 kW	1.0000	396.0 kJ	1.000	15.0
2 × speed module 3	4.00	432 kW	1.2500	345.6 kJ	1.000	12.0
2 × productivity module	1.80	324 kW	0.6075	533.3 kJ	0.926	24.7
2 × productivity module 3	1.40	468 kW	0.5250	891.4 kJ	0.833	28.6
2 × efficiency module	2.00	72 kW	0.6250	115.2 kJ	1.000	24.0

The table reframes when each choice pays off. Speed modules raise throughput per furnace but also raise energy per plate, so they shorten the column at a power cost. A pair of speed module 3s doubles each furnace's output and cuts the column from 24 to 12 furnaces, but every plate costs more energy than a bare steel furnace. This is worthwhile when factory footprint is the binding constraint — when a smelting block needs to fit next to a bus that has no room to grow — and power is plentiful.

Productivity modules go the other way. They slow the furnace down and raise its energy draw, but they produce free plates: a pair of productivity module 3s yields 20% more plates per ore smelted. The energy cost per plate looks terrible in isolation, but the comparison that matters is total ore consumed, not kilojoules per plate. At large scale, saving 16.7% of the ore feed (1 ÷ 1.2 ≈ 0.833) reduces the number of mining drills, the belts feeding the column, and the patch depletion rate. Productivity modules pay when ore is the constraint or when the goal is to minimize raw resource consumption rather than power.

Efficiency modules are the underrated option. Two efficiency module 1s cut the electric furnace's draw to 72 kW — below the steel furnace's 90 kW — while preserving the same speed. That combination gives the electric furnace both lower energy per plate (115.2 kJ vs. 144 kJ) and the pollution reduction that comes with lower power, at zero throughput cost. It does not shorten the column, but if the goal is simply to smelt ore more cheaply than a steel furnace can, this is the breakpoint where the electric furnace first earns its upgrade cost.

When does the electric furnace break even?

The break-even depends on what you are optimizing for. If the metric is energy per plate, a bare electric furnace never catches a steel furnace — it draws twice the power at the same speed. Add two efficiency module 1s and it does catch up, and then some: 72 kW produces the same 0.625 plates per second, beating steel on energy. If the metric is throughput per tile, speed modules are the answer: two speed module 3s double the output of each furnace, which means half as many furnaces for the same belt. If the metric is ore efficiency, productivity modules are the only choice, because nothing else reduces ore consumption.

There is no single "best" module. The common pattern in large bases is productivity modules in furnaces (saving ore at megabase scale) with beacons carrying speed modules to compensate for the speed penalty, but that is a late-game combination. Before beacons, the practical choice is efficiency modules for power-constrained early factories and speed modules when the smelting block needs to shrink.

When the steel 1:1 ratio breaks

The one-to-one ratio is elegant because it needs no arithmetic: one steel furnace per iron furnace, any tier, as long as both furnaces are the same tier. The reason it works is that steel takes exactly 5 times as long as iron and consumes exactly 5 plates per craft, so the consumption rate of one steel furnace matches the production rate of one iron furnace at the same crafting speed.

The ratio breaks the moment the iron and steel furnaces are different tiers. A stone furnace smelts at speed 1; a steel or electric furnace at speed 2. If iron goes into stone furnaces and steel goes into steel furnaces, the steel furnaces consume iron twice as fast as each stone furnace can produce it. The general formula is straightforward:

Iron furnaces per steel furnace = steel furnace speed ÷ iron furnace speed

This is because a steel furnace at speed s consumes iron plates at s ÷ 3.2 per second (5 plates per 16-second craft = 5/16 = 1/3.2 plates per second, times the speed multiplier). An iron furnace at speed v produces v ÷ 3.2 plates per second. The 3.2 cancels, leaving s ÷ v.

Iron furnace	Steel furnace	Iron speed	Steel speed	Iron furnaces per steel furnace	Ratio holds?
stone furnace	stone furnace	1	1	1.0 (exact)	Yes
stone furnace	steel furnace	1	2	2.00	No
stone furnace	electric furnace	1	2	2.00	No
steel furnace	stone furnace	2	1	0.50	No
steel furnace	steel furnace	2	2	1.0 (exact)	Yes
steel furnace	electric furnace	2	2	1.0 (exact)	Yes
electric furnace	stone furnace	2	1	0.50	No
electric furnace	steel furnace	2	2	1.0 (exact)	Yes
electric furnace	electric furnace	2	2	1.0 (exact)	Yes

The table shows that the 1:1 ratio only holds along the diagonal — stone with stone, steel with steel, electric with electric. Every off-diagonal combination deviates. The most common mismatch in practice is iron in electric furnaces (speed 2) feeding steel in stone furnaces (speed 1), which requires two steel furnaces per iron furnace. The reverse — slow iron feeding fast steel — means each steel furnace is starved half the time unless you build twice as many iron furnaces as the 1:1 rule suggests.

This matters because players often upgrade one side of the chain before the other. If you swap iron smelting to electric furnaces but leave steel on steel furnaces, the ratio still holds because both run at speed 2. But if you upgrade steel to electric while iron is still in stone furnaces, you need twice as many iron furnaces. The safe rule is: when in doubt, match the tiers, or calculate using the speed ratio above rather than assuming 1:1.

Why a 24-furnace column can run idle

The 24-furnace column is the textbook answer for a full yellow belt of ore, and it is mathematically correct: 15 ore per second divided by 0.625 plates per furnace per second equals 24. But a correctly sized column in a real factory often has idle furnaces at the far end. The math is not wrong; the assumption behind it is. The calculation assumes ore reaches every furnace. Whether it actually does depends on two things the ratio does not model: inserter throughput and belt compression.

Each furnace pulls ore from the belt with an inserter. Inserters swing on a fixed cycle: they rotate to the belt, pick up items, rotate to the furnace, and drop them. The items per swing and the swing duration determine how many ore per second a single inserter can deliver. If the inserter cannot keep up with the furnace's consumption rate, the furnace waits — even though ore is visible on the belt right next to it. This is usually not a problem with standard yellow inserters feeding basic smelting (the inserter comfortably outpaces a single furnace at speed 2), but it becomes a constraint once speed modules raise the furnace's consumption rate. With two speed module 3s, each furnace pulls 1.25 ore per second, and a single inserter may not sustain that rate depending on the belt tier and inserter type.

Belt compression is the other half. A belt rated for 15 items per second only achieves that rate if it is fully compressed — every lane carries items in every available position. Gaps on the belt reduce effective throughput below the rated maximum. Gaps form wherever items are inserted or removed: a splitter that does not balance evenly, an inserter that places items sporadically, or an underground belt entrance that interrupts the item stream. A column that is mathematically sized for a full belt will be underfed if the belt arrives only 80% compressed. The far-end furnaces see the belt first in the sense that the near-end furnaces consume from it first; by the time the belt reaches the last furnaces, the remaining items may not fill their input slots.

The practical diagnostics are straightforward. If the first few furnaces run continuously but the last few cycle on and off, the belt is losing compression between the source and the column — check splitters and underground belt entrances. If furnaces alternate idle across the whole column (not just at the far end), inserter throughput is the bottleneck, which speed modules can trigger. If every furnace is idle simultaneously, the upstream supply (miners or trains) cannot fill the belt at all, and the column size is irrelevant until the supply is fixed.

These are game mechanics rather than data-file numbers, which is why the ratio table does not model them. The ratio tells you the column length that a perfectly compressed belt with perfectly fast inserters would support. Real factories approach that limit but rarely reach it, and the gap between the theoretical 24 and the practical count is usually one or two idle furnaces, not a design flaw.

Applies when…

• No modules or beacons are fitted. Speed modules shorten the column; productivity modules reduce ore demand instead, which changes the belt side of the calculation rather than the furnace count.
• Furnace counts assume ore reaches every furnace. Inserter throughput and belt compression are separate constraints, and are the usual reason a correctly-sized column has idle furnaces at the far end.
• The one-to-one steel ratio holds when the iron and steel furnaces are the same tier. Mixing tiers breaks it in proportion to the speed difference.
• Energy figures cover furnaces actively smelting. Stone and steel furnaces burn fuel rather than drawing electricity, so their figures are fuel energy rather than grid load.
• These are Nauvis recipes. Casting on Vulcanus produces plates from molten metal and skips smelting entirely.

Related

• Production Ratio Calculator — furnace counts for any target rate
• Belt balancers — keeping every furnace in the column fed
• All ratio pages