Science pack ratios and SPM planning
Science per minute is the yardstick every Factorio base gets measured by, and the number that makes it concrete is rarely stated: how much ore a given SPM target actually consumes, and how many assemblers stand behind it.
Figures computed from game version 2.1.12, updated 2026-08-20.
What one pack per second costs in ore
Expanded through the full recipe tree to raw resources, standard Nauvis recipes, no modules:
| Pack | copper ore | iron ore | Total ore |
|---|---|---|---|
| Red | 1.00 | 2.00 | 3.00 |
| Green | 1.50 | 5.50 | 7.00 |
| Military | 0.50 | 5.75 | 6.25 |
| Blue | 7.50 | 12.0 | 19.5 |
| Purple | 19.2 | 52.5 | 71.7 |
| Yellow | 49.8 | 33.0 | 82.8 |
The spread is the striking part. Yellow science costs 27.6 times the ore of Red science per pack. A base running all six packs at equal rates spends far more of its mining output on the late packs than a glance at the recipes suggests, which is why ore demand climbs so sharply the moment purple and yellow come online.
Assemblers per SPM target
SPM counts packs consumed per minute, so a target of 60 SPM means one pack per second of every type being researched. Assembling machine 3, no modules:
| Pack | Craft time | Yield | 60 SPM | 90 SPM | 1000 SPM |
|---|---|---|---|---|---|
| Red | 5s | 1 | 4.00 | 6.00 | 66.7 |
| Green | 6s | 1 | 4.80 | 7.20 | 80.0 |
| Military | 10s | 2 | 4.00 | 6.00 | 66.7 |
| Blue | 24s | 2 | 9.60 | 14.4 | 160.0 |
| Purple | 21s | 3 | 5.60 | 8.40 | 93.3 |
| Yellow | 21s | 3 | 5.60 | 8.40 | 93.3 |
Craft time alone does not predict the assembler count, because several packs yield more than one per craft. Blue science takes 24 seconds but yields 2, while Purple takes 21 seconds and yields 3. Reading craft time as difficulty gets the ordering wrong.
Does adding labs make research faster?
This is the most common misunderstanding about science, and the answer is no — not in the way people expect.
A lab consumes packs at a fixed rate: one set every research cycle, divided by its speed of 1. Adding a second lab does not make a research finish sooner unless the first lab was idle waiting for packs. What extra labs do is consume more packs in parallel, so research completes faster only when your production can keep them all fed.
Put the other way round: labs are throughput capacity, not a speed multiplier. If your SPM is limited by assembler count or ore supply, adding labs changes nothing at all — they simply spend more time empty. Research speed technology and speed modules in the lab's 2 slots do raise the rate per lab, and those are the levers that matter once the labs are already fed.
The practical check: if the pack buffers feeding your labs stay stocked while research runs, more labs will help. If those buffers sit empty, the bottleneck is upstream and more labs will not change the research rate.
Which science packs share upstream production
The ore table shows what each pack costs in raw resources, but it does not show which packs can share production lines and which need entirely independent supply chains. That distinction matters for layout: two packs that draw on the same ores can pull from a common smelting block, while packs with disjoint inputs need their own dedicated columns.
| Pack | Ore inputs | Other raw inputs | Upstream independence |
|---|---|---|---|
| Red | copper ore, iron ore | none | Mixed ore |
| Green | iron ore, copper ore | none | Mixed ore |
| Military | iron ore, copper ore | coal, stone | Mixed ore |
| Blue | iron ore, copper ore | coal, petroleum gas, water | Requires oil processing |
| Purple | iron ore, copper ore | coal, stone, petroleum gas | Requires oil processing |
| Yellow | iron ore, copper ore | coal, petroleum gas, sulfuric acid, lubricant | Requires oil processing |
Red and green science are the simplest pair: both run entirely on iron and copper. They can share the same smelting columns and belt infrastructure, which is why early-game bases typically place red and green production side by side. Neither requires oil, stone, or coal as a direct input (coal may fuel the miners and furnaces, but it does not appear in the recipe chain).
Military science is the first pack that breaks the iron-and-copper pattern. It requires stone for walls and coal for grenades (through the coal path in the grenade recipe), in addition to iron and copper. The stone and coal demand is small per pack, but it is independent — a base that has been running only red and green may have no stone or coal delivery at all, and adding military science means routing two new resource types to the science block.
Blue science introduces petroleum gas, which is the first fluid input in the chain. This means blue science cannot run without an oil processing setup: refineries, chemical plants for sulfur and plastic, and fluid handling. The iron and copper portions can share existing smelting, but the oil chain is a new dependency that purple and yellow also require.
Purple and yellow science are the heaviest consumers, and they share more upstream than their different recipes suggest. Both need petroleum gas for plastic and advanced circuits; both consume large quantities of iron and copper. The key difference is that yellow science also requires sulfuric acid and lubricant, which means it needs the full oil processing chain plus sulfuric acid production. Purple science requires stone (for stone bricks in the electric furnace recipe) but no additional fluids beyond petroleum gas. The practical implication is that purple and yellow can share their oil and smelting infrastructure, but yellow needs dedicated sulfuric acid and lubricant lines.
No two packs have completely identical upstream requirements, but the grouping for layout purposes is: red and green together (iron/copper only), military on its own branch (adds stone/coal), and blue/purple/yellow together (all require oil). Building the oil block before blue science comes online is the critical path transition — once oil is running, the remaining three packs all draw from it.
Why craft time does not predict assembler count
It is tempting to look at the craft time column and rank packs by difficulty: 24 seconds for blue science sounds harder than 21 seconds for purple. But craft time is only half of the throughput equation. The other half is yield — how many packs come out per craft — and the assembler's crafting speed. The assembler count for one pack per second is:
Assemblers = craft time ÷ (assembler speed × yield)
Blue science takes 24 seconds but yields 2 packs per craft. In an assembling machine 3 at speed 1.25, that is 24 ÷ (1.25 × 2) = 9.6 assemblers for one pack per second. Purple science takes 21 seconds and yields 3 packs: 21 ÷ (1.25 × 3) = 5.6 assemblers. Despite blue having the longer craft time, it needs more assemblers because its yield is lower. Yellow science shares purple's 21-second time and 3-pack yield, landing at the same 5.6 assemblers.
| Pack | Craft time (s) | Yield | Assembler speed | Assemblers per pack/s | Interpretation |
|---|---|---|---|---|---|
| Red | 5 | 1 | 1.25 | 4.00 | One pack per craft — craft time directly sets count |
| Green | 6 | 1 | 1.25 | 4.80 | One pack per craft — craft time directly sets count |
| Military | 10 | 2 | 1.25 | 4.00 | 2 packs per craft offsets the 10s time |
| Blue | 24 | 2 | 1.25 | 9.60 | 2 packs per craft offsets the 24s time |
| Purple | 21 | 3 | 1.25 | 5.60 | 3 packs per craft offsets the 21s time |
| Yellow | 21 | 3 | 1.25 | 5.60 | 3 packs per craft offsets the 21s time |
The general rule for reading any recipe — not just science packs — is that throughput per assembler equals (yield × assembler speed) ÷ craft time. Craft time matters, but a recipe that yields multiple outputs can have high throughput despite a long craft. Blue science is the trap: it has the longest craft time on the table but is not the most assembler-hungry pack per unit of output. Military science, with a 10-second craft and yield of 2, needs 4.00 assemblers — fewer than blue despite being a mid-tier pack, because its yield compensates.
This also explains why upgrading assembler tiers has a different impact across packs. Moving from assembling machine 2 (speed 0.75) to assembling machine 3 (speed 1.25) reduces every count by the same proportion (0.75 ÷ 1.25 = 0.60, about a 40% reduction), but the absolute number of assemblers saved is largest for the packs that already need the most. The assembler count table in the previous section already uses assembling machine 3; downgrading to tier 2 would multiply every count by 1.6666666666666667 and change the physical layout significantly without changing the underlying ratios.
From SPM target to ore production
SPM (science per minute) measures packs consumed, not ore mined. To size the mining operation, you convert the SPM target into a per-second pack rate, multiply by each pack's ore cost, and sum across all six packs. The calculation is one step beyond the tables above: 60 SPM means one pack of each type per second (60 ÷ 60 = 1), and 1000 SPM means 16.67 packs of each type per second (1000 ÷ 60).
| SPM target | Packs/s each type | Total ore/s | Total ore/min | Yellow belts needed | Red belts needed | Electric miners |
|---|---|---|---|---|---|---|
| 60 | 1.00 | 190.3 | 11415 | 12.7 | 6.3 | 381 |
| 90 | 1.50 | 285.4 | 17123 | 19.0 | 9.5 | 571 |
| 1000 | 16.67 | 3170.8 | 190250 | 211.4 | 105.7 | 6342 |
These totals include iron ore, copper ore, coal, and stone — every solid raw material that appears in the six recipe chains. They do not include fluids (petroleum gas, sulfuric acid, lubricant, water), which need their own pumpjack and refinery calculations. The electric miner count assumes a mining drill speed of 0.5 ore per second with no productivity modules; productivity modules in miners reduce the count proportionally.
The per-pack breakdown shows where the ore actually goes. At 60 SPM, red science consumes only 3.0 ore per second, while purple science consumes 71.7 — about 23.9 times as much. The distribution is heavily skewed toward the late-game packs. At 1000 SPM, purple science alone demands 1194 ore per minute, which is more than the entire six-pack total at 60 SPM (11415 per minute).
| Pack | Ore/s at 60 SPM | Ore/s at 90 SPM | Ore/s at 1000 SPM |
|---|---|---|---|
| Red | 3.00 | 4.50 | 50.0 |
| Green | 7.00 | 10.50 | 116.7 |
| Military | 6.25 | 9.38 | 104.2 |
| Blue | 19.50 | 29.25 | 325.0 |
| Purple | 71.67 | 107.50 | 1194.4 |
| Yellow | 82.83 | 124.25 | 1380.6 |
The practical use of these tables is mine planning. A base targeting 60 SPM needs roughly 381 electric miners across all ore types, which is a modest outpost. At 1000 SPM, the same calculation gives 6342 miners — a number that usually requires multiple outposts, rail delivery, and productivity modules to keep the drill count manageable. The belt counts are equally telling: 60 SPM fits on 12.7 yellow belts of total ore, while 1000 SPM needs 105.7 red belts or 70.5 blue belts just for raw ore delivery, before any smelting or intermediate production.
The key planning insight is that ore demand does not scale linearly with the number of packs — it scales with which packs are online. A base running only red and green at 60 SPM uses a fraction of the total. Adding military adds stone and coal. Adding blue, purple, and yellow introduces oil and multiplies iron and copper demand several times over. The SPM target alone does not tell you how much ore you need; the pack mix does.
How research selection changes real ore demand
The tables above assume every pack type is consumed at the full target rate. In practice, research technologies consume different subsets of packs, and the actual ore demand depends on which technology is being researched at any given moment. Red and green science are consumed by nearly every early-game technology, so their production runs continuously. Military science is only consumed by military and late-game technologies, so a base that is researching purely industrial technologies may produce military packs that sit in buffers until a military technology comes up.
This matters for two reasons. First, building all six pack types at exactly the SPM target wastes ore on packs that are not currently being consumed, unless the goal is to maintain buffer stock. Second, the lab count calculation interacts with this: labs consume one of every required pack per research cycle, so a technology that requires only three pack types consumes those three at full rate but does not touch the other three at all. The practical approach is to size each pack's production for its average consumption rate across the technologies being researched, not for the theoretical maximum. A base that alternates between military and industrial research can run military science at a lower sustained rate without stalling.
One additional consideration is that productivity modules in research labs do not exist in the same form as assembler productivity — labs accept speed modules, not productivity modules, so the only way to consume fewer packs per research point is through research speed technologies. Those technologies reduce the number of cycles needed per technology but do not change the per-cycle pack consumption. The ore cost per technology is therefore fixed by the recipe totals; speed modules and lab count determine how fast that ore is consumed, not how much is consumed in total.
Applies when…
- No modules or beacons are fitted. Productivity modules in the pack assemblers reduce ore demand; speed modules reduce assembler counts.
- Recipes are the standard Nauvis ones. The Space Age packs made on other planets have their own chains and are not included in these totals.
- Fluid inputs are treated as raw and not expanded further, so packs needing petroleum gas or sulfuric acid have costs beyond the ore columns shown.
- SPM figures assume every pack type is consumed at the target rate. A research that does not use military science needs no military packs, and the ore total drops accordingly.
- Assembler counts are exact fractions. Rounding down misses the target rate; rounding up leaves idle capacity.
Related
- Production Ratio Calculator — full machine counts for any pack and rate
- Smelting ratios — the furnace columns feeding all of this
- All ratio pages