SPM Atlas

Smelting layouts: columns, belt feeds and beaconed blocks

A smelting layout is usually shared as a blueprint to place. The dimensions behind it are a small amount of arithmetic: how fast each furnace crafts, how much a belt carries, and how many furnaces fit between the two belts that feed and drain them. Once those numbers are on the page, the layout can be sized for any belt tier or target throughput instead of copied from somewhere else.

Smelting figures computed from game version 2.1.12, updated 2026-08-22.

The three furnaces

Stone, steel and electric furnaces differ in three ways that matter for layout: crafting speed, power draw, and module slots. The speed decides how many furnaces a column needs; the power decides what fuels or supplies it; the slots decide whether beacons and modules can do anything for it at all.

Furnace Crafting speed Power draw Module slots Iron plates / s kJ / plate
stone furnace 1 90 kW 0 0.31 288
steel furnace 2 90 kW 0 0.63 144
electric furnace 2 180 kW 2 0.63 288

Iron plate takes 3.2 seconds per craft. A stone furnace at speed 1 turns out 0.31 plates per second; steel and electric furnaces both run at speed 2 and turn out twice that. The energy used per plate is what actually separates them: a steel furnace burns fuel for its 144 kJ per plate while an electric furnace uses grid power for 288 kJ per plate at the same output. Stone and steel furnaces have no module slots; only the electric furnace can receive modules or beacons.

How a belt-fed column is sized

A full belt of ore feeds a row of furnaces, each taking one ore and producing one plate. If every furnace is to be kept running, the number of furnaces in the column is the belt's item rate divided by each furnace's output rate. With iron plate at 3.2 seconds and an electric furnace at speed 2, one furnace produces 0.63 plates per second, so a yellow belt at 15.0 items per second needs 24.0 furnaces to consume it completely.

The same arithmetic applies at every belt tier and for every furnace. Stone furnaces craft at half the speed of the other two, so a full-belt stone column needs twice as many furnaces as an equivalent steel or electric column.

Belt in Items / s Stone furnaces Steel furnaces Electric furnaces
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 72.0 72.0
turbo transport belt 60.0 192 96.0 96.0

These counts are identical for stone brick, which also takes 3.2 seconds per craft and produces one brick from two stone. The recipe consumes two items per craft rather than one, so the input belt is drained in half the distance and the column can be half as long if the only goal is to consume the feed; the output belt of bricks, however, still carries at half the input item rate because two stone become one brick. A full output belt of bricks therefore needs two belts of stone feeding in.

The basic column shape

The arrangement that gives those numbers room is an input belt along one side, an output belt along the other, and furnaces in between. Inserters face across the gap: one pulls ore from the input belt into a furnace, the other pulls plates out onto the output belt. 

  input belt  →  [F][F][F][F][F][F] ... [F]  →  output belt
                   |  |  |  |  |  |         |
              (ore in, plates out, one inserter each side)

The column's length is set by the furnace count in the table above. Its width is set by what stands between the belts: the furnaces themselves, the inserters on each side, and whatever room the belts need to turn around at the ends. In practice a furnace column includes a turnaround tile at each end so the input and output belts do not have to be threaded through the row. That turnaround is part of the footprint but not part of the furnace count.

A second output belt can be added opposite the first so that plates leave on both sides. This does not change the furnace count; it halves the load on each output belt, which only matters once a single output belt cannot carry the plates the column produces. For a yellow belt feeding 24.0 electric furnaces, the output rate equals the input rate (one plate per ore), so one output belt is exactly enough.

Steel: the column that consumes five iron belts

Steel plate is not a direct smelt of ore. It takes 5 iron plates and 16 seconds to craft one steel plate in a furnace. That has two consequences for a column layout. First, the furnaces that make steel are slow — 16 seconds versus 3.2 for iron — so an output belt of steel needs many more of them than an output belt of iron. Second, those furnaces must be fed iron plates, which themselves need furnaces upstream.

To saturate one yellow output belt of steel at 15.0 plates per second, the steel furnaces must consume 75.0 iron plates per second — equal to 5.00 yellow belts of iron. The two-stage column therefore has a row of steel furnaces fed by a small iron-smelting block large enough to supply it.

Output belt Steel out / s Iron in / s Yellow belts of iron Steel furnaces Iron furnaces
transport belt 15.0 75.0 5.00 120 120
fast transport belt 30.0 150 10.0 240 240
express transport belt 45.0 225 15.0 360 360
turbo transport belt 60.0 300 20.0 480 480

The shared solver confirms the same totals: at one yellow belt of steel output the top stage reports 120 electric furnaces for steel and 120 for the iron plates that feed them. The "five to one" ratio is not a coincidence of the belt tier; it holds at every tier because 16 divided by 3.2 equals 5, which is exactly the iron-per-craft number. A single yellow belt of steel therefore always demands five yellow belts of iron plate.

When to switch from stone to steel to electric

The furnace table shows that steel and electric furnaces craft at the same speed (2) while stone furnaces craft at half that (1). Speed alone therefore never decides between steel and electric; it decides between stone and everything else.

Stone furnaces are the early-game option because they cost stone and burn coal directly. A full yellow belt of ore needs 48.0 stone furnaces — twice the count of a steel or electric column. That footprint is the real cost once ore throughput grows: a yellow-belt stone column is twice as long as the equivalent electric one, and a red-belt stone column reaches 96.0 furnaces, which is long enough to make routing around it awkward.

Steel furnaces are the straightforward upgrade: same speed as electric, no power network required, fuelled by coal or solid fuel. They halve the furnace count versus stone without adding any electrical infrastructure. Their cost is steel, which is itself a smelting product, so bootstrapping the first steel furnace column usually means smelting the steel it is built from in stone furnaces first.

Electric furnaces match steel furnaces on speed but draw 180 kW from the grid instead of burning fuel. That is worth doing for three reasons that matter to layout, not just to fuel logistics. Electric furnaces accept modules, which is the only way to shrink a column's footprint with productivity or speed effects. They can be beaconed, which is the only way to reach the high throughputs a megabase needs without laying down hundreds of furnaces per belt. And they do not need fuel inserters or coal belts, which removes a supply line from each column.

The practical switch points follow from those differences. Switch from stone to steel when the stone column for a full red belt (96.0 furnaces) is too long to route around, or when the coal to fuel it is competing with plastic and explosives. Switch from steel to electric when you have modules to insert (productivity modules pay for themselves in ore saved), when you want to beacon a block, or when the coal belts feeding every column are more trouble than a power trunk line.

Beaconed smelting blocks

A beacon transmits module effects to machines within its supply area of 3 tiles. Each beacon holds two modules; in version 2.1.12 each module's effect is multiplied by a distribution effectivity of 1.5 before being applied, so a single beacon with two speed module 3 delivers a total speed effect of 150% to every furnace it covers. A furnace inside range of several beacons receives the sum of all of them.

The layout consequence is that furnaces stop being arranged as one long column and start being arranged as a compact block ringed by beacons. A furnace in the interior of an 8-beacon block sees a speed effect of 1200%, which raises its crafting speed from 2 to 26.0. The number of furnaces needed to consume a yellow belt drops from 24.0 to 1.85.

Configuration Speed effect Effective speed Furnaces per yellow belt
No beacons, no modules 0% 2.00 24.0
1 beacon (2 × speed module 3) 150% 5.00 9.60
8 beacons (each 2 × speed module 3) 1200% 26.0 1.85

The beaconed block is smaller in furnace count but larger in power draw. Each beacon consumes 480 kW continuously, regardless of whether the furnaces it covers are working, and speed modules raise the furnaces' own power draw. A block ringed by eight beacons pays eight beacon power draws on top of the faster-burning furnaces; the trade-off is footprint and ore throughput per tile, not energy per plate. Productivity modules in the furnace slots change the calculus further: they reduce ore consumption per plate at the cost of speed, which is usually worth more in a megabase than the raw speed bonus from filling the beacon ring with speed modules.

One layout point is worth making explicit. A beacon's supply area is a square, not a circle, and beacons must be within range of the furnaces they affect rather than merely nearby. A row of beacons down each long side of a compact furnace block covers two rows of furnaces each; a beacon at each corner covers the end furnaces that the side beacons miss. The exact number of beacons a block needs is set by how many furnaces are in the interior and how far the supply area reaches, which is why beaconed blocks are planned as rectangles rather than as arbitrary shapes.

Comparing belt-fed columns across tiers

The furnace-count table above makes the belt-tier comparison direct. A turbo belt carries 60.0 items per second, four times a yellow belt. An electric-furnace column sized for a turbo belt therefore needs 96.0 furnaces — four times the count of a yellow column in exactly the same footprint ratio. If a yellow column of 24.0 furnaces is already a long row, a turbo column of 96.0 is long enough that most bases split it into several shorter columns fed by splitters rather than running one unbroken line.

Splitting a full belt across two columns is a balancer problem: the feed must divide evenly so neither column is starved while the other is overfed. A two-way split with a single splitter is sufficient for two columns; four columns need a 4-to-4 balancer. The balancer design guide covers the splitter arithmetic; the point here is that the column count and the balancer width are chosen together.

Target throughput Yellow columns Red columns Blue columns Turbo columns
15.0 / s 1 1 1 1
30.0 / s 2 1 1 1
45.0 / s 3 2 1 1
60.0 / s 4 2 2 1

These are the number of full single-belt columns, not the number of belts. Three yellow columns carry the same plates per second as one blue column, but they use three times the footprint and three times the inserters. The usual rule is to use the fastest belt you can supply power and logistics for, and split into multiple columns only when a single column's length becomes awkward to route around.

What the solver reports for a full belt

Running the shared ratio solver for a yellow belt of iron plates returns one stage: 24.0 electric furnaces, with iron ore as the raw input at 15.0 per second. There are no intermediate products because iron plate is a direct smelt. Running it for a yellow belt of steel returns two stages: 120 furnaces for steel and 120 for the iron feeding them, with 75.0 iron ore per second as the raw input.

Those totals are what a layout must provide. Everything else — the belt direction, the inserter facing, the turnaround at the ends, the beacon spacing — is geometry around those counts. The reason to understand the counts rather than copy a blueprint is that the counts change with belt tier, furnace tier, modules and target SPM, and a fixed design cannot track all four at once.

Applies when…

  • Furnace counts assume one ore in, one plate out for iron and copper. Stone brick consumes two stone per brick, so its input belt drains twice as fast as its output belt fills.
  • Belt throughputs assume both lanes are full. A single-lane feed carries half the stated rate and needs half the furnaces to consume it.
  • Beacon figures use two speed module 3 per beacon and a supply area of 3 tiles from machines.json. Productivity modules and quality modules change both the count and the ore demand and are not modelled in the comparison table.
  • Power per plate is measured at the furnace's rated draw without module penalties. Speed modules raise draw; productivity modules raise it further; efficiency modules can reduce it.
  • Steel ratios assume iron plate is smelted on-site in electric furnaces at the same speed. Feeding steel from an external bus changes the upstream furnace count but not the steel furnace count.

Related