Factorio Science Pack Ratios Guide: Red, Green & Blue Science Production

Red science is the first place to learn clean ratio thinking because the recipe is short and the bottlenecks are obvious. The literal recipe is 1 iron gear wheel + 1 copper plate per automation science pack. In quick planning shorthand, many players treat the line as 0.5 iron gear + 0.5 copper plate per science step because the block is split evenly between one gear input and one plate input.

The assembler math is what matters for a blueprint. A gear wheel recipe is so fast that one gear assembler can support roughly ten red science assemblers at base machine speeds. That means the clean starter blueprint is not one gear machine for every science machine. It is a compact block with a single gear assembler feeding a longer row of pack assemblers, while copper is delivered directly on a belt with no intermediate crafting step.

The usual mistake is overbuilding gears and underfeeding copper. If the gear chest is full but science stalls, your problem is not assembler count. Your copper lane is starving or your inserters are not keeping up with the belt spacing. Red science should feel boring, and if it does not, the blueprint is already telling you something upstream is wrong.

Green science is where simple science lines start branching. Each logistic science pack needs one inserter and one transport belt, but those two ingredients do not cost the same to produce. Inserters pull in gears, circuits, and iron plates, while belts are mostly cheap iron and gears with a fast output recipe. The result is that inserters, not belts, become the common choke point.

A practical green science blueprint uses a shared gear feed, a small belt sub-line, and a larger inserter sub-line. A stable ratio block is 1 belt assembler : 2 inserter assemblers : 24 logistic science assemblers. You can halve or quarter that for smaller bases, but the relationship stays the same: inserters deserve more machine budget than belts.

Common bottlenecks are predictable. Circuits starve the inserter line first, gears get siphoned away by belts and inserters at the same time, and mixed output belts become messy if inserters are crossing lanes inefficiently. If your green science automation looks inconsistent, start by checking the inserter recipe chain before adding more science assemblers.

Blue science is the first science pack that forces your base to behave like an integrated system. Chemical science packs need engine units, advanced circuits, and sulfur, so one missing fluid condition can idle the whole block. That is why Factorio blue science production is less about one isolated assembler ratio and more about whether the refinery and circuit layers are already stable.

A reliable planning ratio is 5 engine assemblers : 6 advanced circuit assemblers : 12 blue science assemblers, with sulfur supplied from a shared oil block. That keeps the engine side and advanced circuit side close enough that neither idles the other while the oil network handles petroleum and sulfur buffering separately.

The real bottleneck is rarely the final science assembler. It is usually oil instability, plastic shortages feeding red circuits, or pipe backups that make sulfur intermittent. When players search for a blue science production blueprint, they usually need a pre-balanced chain that already includes those dependencies instead of another partial fix at the last assembler row.

Military science and utility science do not belong in the same early-game block, but they should be planned with the same discipline. Military science is item-heavy and tends to stall on grenades or piercing rounds. Utility science is infrastructure-heavy and usually bottlenecks on processing units or flying robot frames. In both cases, the slowest subcomponent defines the effective science-per-minute rate of the whole line.

The takeaway is consistent across all colors: size the expensive ingredient chain first, then let the final science assemblers reflect that ceiling. A science pack blueprint that only counts the last row of assemblers is not actually balanced.

Labs only move as fast as the slowest science color feeding them. The clean way to scale research is to choose a target science-per-minute number first, then size every science block to that same target. If red can deliver 120 SPM but blue can only hold 45 SPM, your labs are a 45-SPM design no matter how healthy the early science belts look.

Use mirrored lab inputs, equal belt lengths where possible, and small buffers instead of giant chest stockpiles. Large buffers hide ratio mistakes. Small buffers expose them quickly, which is exactly what you want when tuning a research mall. Once every color hits the same sustained rate, adding more labs or lab speed becomes a clean scale-up instead of a debugging session.

AI agents are bad at improvising around hidden ratio debt. If a science block depends on an undocumented inserter shortage, fluid backup, or lane-mixing trick, the agent has to rediscover the failure mode from observation. That is slow and unreliable.

A pre-balanced science pack blueprint removes that uncertainty. Machine counts, belt lanes, and expected outputs are already locked. The agent can place the block, verify the deterministic layout, and move on to the next planning step without guessing whether the science line is only temporarily stable.