How to Choose Your Factorio Factory Design Pattern - 3 Key Criteria

Around the point where you're progressing through red to blue science, your factory can suddenly feel cramped. I hit the wall with main bus width constraints and the yellow belt's 15 items/second ceiling, forcing a complete rebuild.

This article focuses on Factorio vanilla 2.0 as the primary baseline, while also considering how design premises shift with Space Age. I'll walk intermediate players through four patterns—spaghetti, main bus, modular, and train-based—to help you understand which suits your factory best.

The key isn't memorising "the perfect layout," but rather judging your factory against three axes: throughput demand, available space, and future expansion method. Which materials should you transport by belt over distance, and which should you shift to local production? Using the belt, beacon, rail, and power values available, I'll explain the actual decision process in concrete terms.

The main focus here is vanilla 2.0. The design decision foundation is based on belt transport, rail, beacons, and power. I stick to confirmed figures. For example, yellow transport belt throughput is 15 items/second. Red and blue belt speeds are calculated as multiples relative to yellow (red = 2×, blue = 3×), so in theory red ≈ 30 items/s and blue ≈ 45 items/s (calculated as "15 items/s × multiplier"). I treat these as guidance here; if a primary source exists for absolute red/blue values, I cite that instead. When setting main bus width, I use this conversion to decide "should I double-track it later, or split onto separate systems from the start?"

Main Bus consolidates high-demand materials—iron plates, copper plates, electronic circuits—into trunk belt lines, with production branches splitting sideways to draw them. This is framed as a structuring philosophy. Beginners favour it because flow direction locks in, making "source → sink" tracking much easier than spaghetti sprawl. The cost: trunk belt count and footprint run large. Push wrong, and you get a factory that's "tidy but stretches to the horizon." Best for: red-green onward → blue science band, mid-game workhorse.

What to Do When Your Factory Feels Cramped

How This Article Is Structured and What to Look For

This article doesn't decide factory design by asking "which is strongest?" Instead, it asks: what's your current bottleneck? Are you rushing early progress, did your layout break down in the blue science phase, or do you want to extract resources via train in the late game? The answer changes which approach you should choose. I'll lay out all four patterns—spaghetti, main bus, modular, and train-based—so you can match your specific problem to the right one.

Let me align terminology first. The Main Bus is a concept where high-frequency materials like iron plates, copper plates, and electronic circuits are consolidated onto trunk belt lines, with each production line branching sideways to draw from them. It's often treated as a beginner-friendly base type because it's easy to structure. The strength lies in how it organises your factory. However, because it uses considerable belt space and horizontal width, poor expansion technique can lead to a factory that "looks tidy but stretches forever."

Beacons are buildings that spread module effects across a 9×9 area at half strength. They're critical for density, but consume 480kW each, so placing more means heavier power planning. They're not early-game heroes—they shine late when you're asking "how much can I pack into the same footprint?" Stackers (station wait areas) are queuing lanes you build in front of train stations. Without them in a multi-train design, congestion and choking worsen dramatically.

UPS is essentially how the game feels when your factory becomes massive—a reflection of processing load. Design quality shows not just in output, but in late-game smoothness. It ties to your play philosophy: how big do you want to push the factory? Quality, added in Space Age, brings a different mindset. Instead of just expanding your factory horizontally, you can now strengthen equipment and machines vertically through higher quality tiers. This concept shifts design assumptions away from vanilla 2.0 alone.

On my first clear, I just dropped assemblers and belts wherever there was free space, hit blue science, and completely lost track. Was iron short, or were circuits backed up? Where do I even start fixing this? That's the classic breakdown. From there, I stopped winging layouts and switched to "put this decision table beside me, declare 'main bus up to this point, then switch to trains,' and commit to it." Factory design depends less on intuition and much more on having clear switching conditions.

The comparison logic is simple too. For speed of setup, spaghetti still wins. For stability to first clear, main bus is easier. For late-game scaling and expansion, modular or train-based pulls ahead. The official Tutorial mentions main bus as beginner-friendly, but that doesn't mean "always optimal"—it means uniquely good as a low-failure entry point. That's the most honest way to frame it.

One detail worth emphasizing: transport methods split into three branches—belts, rails, and logistic robots. Every design pattern's character is set by which one you make the star. Main bus leans belt-heavy, train-based leans rail-heavy, and some late modules lean robot-heavy. If you frame pattern comparison as "which transport mode becomes primary?" rather than visual layout, it becomes much clearer to read.

Tutorials

wiki.factorio.com

Target Version and Where the Numbers Come From

The main focus here is vanilla 2.0. The design decision foundation is based on official Wiki entries for belt transport, rail, beacons, and power. I stick to confirmed figures. For example, yellow transport belt throughput is 15 items/second per official records. Red and blue belt speeds are calculated as multiples relative to yellow (red = 2×, blue = 3×), so in theory red ≈ 30 items/s and blue ≈ 45 items/s (calculated as "15 items/s × multiplier"). I treat these as guidance here; if a primary source exists for absolute red/blue values, I cite that instead. When setting main bus width, I use this conversion to decide "should I double-track it later, or split onto separate systems from the start?"

Beacons matter just as much for late-stage design. A beacon spreads effect across 9×9, with half transfer rate. It costs 480kW though, so "the more you place, the more power planning weighs on you." Looking at the ratio, beacon factories don't just save space—they front-load the burden elsewhere (power, area, wiring complexity). This mental model changes how you approach them.

Power numbers serve as a design calibration tool too. Optimal solar-to-accumulator ratio is 21:25, and supporting 1MW day and night requires 23.8 solar panels. While this article's focus is factory layout, the further you lean into beacons or logistics networks, the more power planning becomes part of layout itself. Factory design and power design aren't separate problems—they're deeply linked.

For rail systems, signal and stacker handling is the decision lever. Train networks use signals to partition track into blocks, fundamentally supporting multi-train operation. If multiple trains converge on one station, a stacker in front prevents gridlock. Train-based gets called "end-game friendly" not just for distance, but because you can systematize expansion through stations, signals, and stackers. Flip side: jump to train-based too early without mastering the system, and breakdowns become harder to debug than belt factories.

Let me set Space Age's position clearly. It's a paid DLC released 21 October 2024, adding four new planets, space platforms, and restructured tech trees. Three major additions are Space Age itself, Quality, and Elevated Rails, expanding beyond the single-ground-factory-stretching-sideways model. Vertical upgrade via quality means "multiply the line by 2" isn't your only answer anymore.

Because of this, I split the content into established facts and practical design judgement flowing from them. "Main bus suits beginners," "beacons enable density," "multi-train needs signals and stackers"—these are fact-based. "For first clear, main bus stabilizes easier," "station modularity beats single-base design"—these are interpretation grounded in official specs. No design pattern has an official "one true answer," so I keep that separation clear throughout.

Old Steam guides and community posts sometimes mix versions. They're useful as supplementary thinking aids but not the foundation for this article's numbers and specs. My comparison tables assume vanilla 2.0 factory growth, asking which pattern fits which mid-game concern. With that framing, the question "why does that method suit that phase?" becomes much easier to answer.

Four Design Patterns: The Core Models

Defining Each Pattern and Its Scale

Factory design gets confusing because appearance and logistics get tangled together. I'll cut the four patterns by "what becomes your primary lever for scaling?" This sorting helped me judge, when stumped at blue science, whether a simple add-on would fix it or whether the whole system needed rethinking.

Spaghetti means chaining whatever you need next to whatever empty space you find. Setup speed is fastest of the four, strong through red and green science. Belt lengths stay short, perfect if you're racing through early tech while chewing through a starter ore patch. Later, adding iron lines or circuit lines creates crossings, and visibility crashes. You lose track of what feeds what and where the real bottleneck is. Best for: startup through red/green breakthrough.

Main Bus consolidates high-demand materials—iron plates, copper plates, electronic circuits—into trunk belt lines, with production branches splitting sideways to draw them. The Wiki's Main Bus article frames this as a structuring philosophy. Beginners favour it because flow direction locks in, making "source → sink" tracking much easier than spaghetti sprawl. The cost: trunk belt count and footprint run large. Push wrong, and you get a factory that's "tidy but stretches to the horizon." Best for: red-green onward → blue science band, mid-game workhorse.

Modular splits production by function—electronic board factory, smelting block, science block—into independent compartments. The win is that you can copy and scale up one "unit" rather than retuning a single line. Lock the compartment's input/output, and you expand by duplicating. Downside: you must pre-decide what each block contains, what it makes locally, what it imports. High expansion potential but high design upfront cost. Best for: blue onward where complexity balloons through endgame.

Train-Based (Distributed) spreads smelting, circuits, oil, science across distant outposts, linking them with stations and trains. Shines for long-haul and mega-scale; if ore patches run out, your whole factory doesn't need rebuilding. The catch: stations and signals become your logistics foundation. Per the Rails wiki, trains need signals to partition track into blocks before multi-train safety works. Multiple trains at one station need stackers to dodge pile-ups. Best for: late-game big-picture expansion. Workable earlier, but signal understanding gaps make debugging messier than belts.

In practice, phases overlap. My gut call: run spaghetti through red-green, jump to main bus around blue, then modularise production blocks, then train-ify when needed. Least accident risk, lowest rebuild cost.

User:Fried biter/workspace2

wiki.factorio.com

Comparison Table

Words blur; let me tabulate by asking "what worry does each method handle well?"—not "which is strongest?"

The standout tension: expansion vs. upfront planning. Spaghetti kills today's research fastest, but tomorrow's scaling hurts. Train-based handles tomorrow perfectly, but today it's heavy. Main bus splits the difference—it gives beginners enough order to stay stable mid-run, making it a natural waypoint.

Confusing modular and train-based as both "expansion-friendly" trips people up. The difference: modular's hero is compartment-splitting (transport can be belt or robot). Train-based's hero is the transport itself, scattering factories around stations. Want to chop production into pieces, or connect pieces over distance? Different answers.

Decision Flowchart

Can't pick? Use phase + ache to narrow fast. Design fits what hurts now, not preference.

1. Red–Green science, maxing research and grabber speed

Spaghetti dominates here. Furnace, power, defence, early labs—their speed beats cosmetics.

1. Blue science landed, iron-circuit-oil traces vanish

Time to jump main bus. One trunk-line pass makes flow and shortage instantly visible. Most new players get factory design for the first time here.

1. Main bus running, but specific production batches beg independence

Modular clicks here. Circuit factory, oil block, science batch—cut it as a self-contained unit. Repair scope shrinks; swap the problem block rather than retool everything.

1. Ore sites distant, multiple mega-blocks spinning together

Train-based shows up. Belt-marathons fade; standardised loading stations, hubs, and main lines feel more stable. Multi-train operation needs signal and stacker discipline.

1. Space Age: inter-planet trade and quality upgrades on your mind

Single mega-bus loses appeal; planet-by-planet blocks + careful export mix fit better. Quality means stacking isn't just volume—it's tiers.

Compressed: rush → spaghetti. Organize → main bus. Replicate → modular. Scatter → train. Lock these four in first; then deep reading gets easier.

Three Criteria for Picking Your Factory Design

Three axes snap the four patterns into focus. I judge design not by looks but by how much you can add later, how legibly problems show up, and how much you nail down upfront. Factories spend more time expanding than perfect. These three angles catch failures earlier.

Criterion 1: Expansion Measured in Numbers

"Is it easy to grow?" isn't a feeling—it's space left to widen, blocks that scale independently, transport you can later decentralise? Spaghetti chokes because free squares exist but wiring ties them down. Main bus is built for "add another line beside the trunk"—blank space = viable expansion. Modular goes further: duplicate entire circuit blocks or oil zones wholesale.

Train-based's superpower: push growth outside the same footprint. Copper smelter 2, circuit factory B, science base 2—all on separate stations. You're not cramming the same land; you're fanning out. This dispersal headroom scales with distance.

Belt ceilings anchor the math. Yellow belt maxes at 15 items/sec. Running single-belt long-distance for a heavy-demand mid-tier? Dead fast. My blue science expansion tanked dragging copper wire one line—instant shortage. Switching to "make wire local to circuit use" unlocked everything. Expansion sense flips: don't ask "more belts?"—ask "can we stop transporting this?" That shift is game-changing.

Criterion 2: Seeing Where It Breaks

Good sight-lines mean you hunt problems fast. Main bus mid-game shines because you count trunk belts by eye: "iron ×4, copper ×4, circuits ×2," capacity's obvious, "not enough production" vs. "bottleneck in distribution" separate cleanly. Spaghetti: fingers point everywhere; you extend belts to patch leaks forever without finding root cause.

The 15 items/sec yellow belt is your diagnostic compass. If your design runs a single-belt long-haul for heavy consumption, you will saturate fast. Low-sight designs band-aid with more belts; high-sight designs notice "supply's fine, distribution broke" and redesign.

Train networks need visibility too, except the failure mode is sneakier. Trains hide jams better than belt tangles. So role-splitting matters: "this corridor is a road, this zone is a factory, here signals partition blocks, here a station waits." Rails sit on 2-tile increments, so sloppy positioning locks away future flexibility. Just as belt-count logic applies to the trunk, track geometry logic applies to the frame. Patterns that let you see "passage" vs. "work zone" at a glance stay strong longer.

Belt transport system

wiki.factorio.com

Criterion 3: How Much You Plan Upfront

How much scaffolding do you need to pour before you start? This is the most-skipped decision lever. Spaghetti lets you build now, ceding stability to the future you. Main bus needs trunk width and grow-direction nailed; then go. Modular and train-based demand stations, signals, power, land reserves—fail here, they crumble.

Train design shows this starkest. Multi-train shared stations need stackers before the station, sequestering waiting trains off the main line. Tutorials: Train Signals emphasizes this—dead-lock avoidance means waiting outside the main track. Stations feeling congested isn't a train shortage; it's no staging area. One set of "station + wait space" counts as a module. Train-based only "works" when station design includes staging.

Power's the same. Loose designs improvise; scaled ones calculate. Solar: use the 21:25 ratio, remember 23.8 panels per 1MW sustained, and reserve land. Modular and train-based benefit hugely from pre-planning here. The payoff appears late: "I added this block, and nothing brownouts" feels like magic until you realize upfront electricity budgeting made it routine.

This lens splits patterns sharply. Low upfront-plan tolerance? Spaghetti or main bus. Want standards set, then autopilot? Modular or train-based. Build quality = how straight expansion runs, not how pretty the blueprint is.

Power production

wiki.factorio.com

Why Main Bus Is the Beginner Recommendation

Main Bus Fundamentals

I keep recommending main bus to newcomers. It consolidates high-frequency materials into parallel trunk lines, splitting sideways branches to feed production. Think of it as the factory's central spine, then grow sideways from there. The payoff: tracking problems is much easier.

Why beginner-friendly? Debug-ability beats aesthetics. Spaghetti tangles blame: short on iron, wiring crossed, or another line stealing? All knotted. Main bus? Count trunk belts visually—"iron ×4, copper ×4, circuit ×2"—spot thinning, cut the problem from that. I felt this relief firsthand: splitting "flow path" from "work area" alone sharpened thinking a tier.

Layout basics stay simple. Tier main materials vertically, flank one side with production, expanding sideways. Need a new good? Branch a sideroad from the trunk, build sideways. The ethos: trunk carries; wings process. This splits your mental model neatly.

Scope's also clean: main bus dominates blue science. Mid-run, ingredient variety stays manageable, widening the trunk scales smoothly. Beyond that, trunk diversity and branch count explode; sustaining a single mega-bus costs real management overhead. Late-stage you're constantly asking "trunk this item, or make local?" which makes main bus a phase tool, not an eternal truth.

The drawback stings: massive space hunger. Post-widening is awkward, so reserve lanes upfront or get choked. My first attempt burned through margins by hoarding width. Second time, I pre-marked 4 empty lanes and made clear of blue science. Breathing room beats density.

What To Run, What To Stop

Newcomers itch to "flow everything central"—stop there. Basis: high-frequency, multi-use, multi-consumer materials. Iron and copper plates? Classic trunk riders. Electronic circuits? Broad enough for bussing. That logic: prioritise items used everywhere, continuously, from many angles.

Flip side: don't rush low-density bulk. Copper wire hits consumption hard but bloats in volume. It collapses the trunk visually, line-counts spike, organization crumbles. Shipping copper plate to wire sites lets you convert locally—cleaner, faster.

That 15 items/sec yellow belt cap drives the call. Blind "fit it all on the trunk" saturates demand-heavy lines first, then you band-aid forever. Main bus isn't "pile everything central"; it's choiceful curation. Flow foundation resources, local-produce mid-tier balloons. Long-lived buses starve volume greed.

Practical shorthand: ship plates, make swollen mid-tiers locally. Iron/copper plates worth bussing; circuits expand so much on-site beats long haul. Choose this, and bus life extends markedly.

Cheat Sheet: Bus vs. Make Local

Overthinking this call every cycle burns energy. My go-to split:

Logic: broad, multi-way needs = bus friendly. Bulky, single-use = make-local friendly. Main bus beginner play nails blue if you starve the trunk of balloons. That discipline extends runway hugely. Deeper concept in modular primers; for now, "keep trunk lean" suffices.

When to Jump to Modular or Train-Based Mid-Run

The Handoff Trigger

Main bus pain shows visibly: the factory stretches into eternity. Each new item spawns another trunk line, another empty buffer, another production wing. Needed zones drift further; expansion itself becomes logistics cost. More trunk lanes don't clarify state—they obscure it. Visibility drops, bottlenecks blur.

Second pain point: trunk saturation and downstream starvation. Yellow belt hits 15 items/sec hard. Mid-game piles demand—circuits, steel, compound—and even more lines can't dodge the structure's fundamental "pour everything to center" ceiling. Here, splitting production into independent blocks that feed themselves beats fattening the bus.

My switch signal: upgrading mid-tier goods—reds, blues—feels easier in a dedicated zone than crammed back into the main bus spine. That's the moment to isolate production by output, let each clone independently. Modular is slower to start but exponentially easier to grow.

This primes train-base too. Long-haul plates and ore cost belts fast; trains carry distance cheaper. Split smelting, circuits, science across sites, train between them: trunk ×1 becomes hub-and-spoke. Stability rocketed when I framed it as "train ferries bulk goods; belts close loops inside each block." Late-scale train beats mega-bus every time.

Beacon-Ready Block Sizes

Jumping modular requires pre-sizing blocks for beacon endgame. Beacons span 9×9, half effect, but 480kW burn each. Underbudget them, rebuild it all late. Front-load here, and flexibility reigns.

I cut zones by product—red circuit block, blue circuit block, oil block—each duplicable horizontally. Need more? Clone the blueprint, plop it down. Main bus agony—"where on the trunk do I tap?"—vanishes. Now it's just "how many blocks?" Ratios lock in.

Grid discipline shines if rails enter play. Tracks sit on 2-tile steps, so 1-tile finesse fails. City-block patterns win by standardizing block×rail×station grid upfront. Future expansion flows on rails instead of improvised belts. Late-game that pre-planning is the difference between "we can add a 10th block" and "the whole thing's too tangled."

Power budgets collide hard here. Dense beacon zones drink power: 480kW × beacon count. Rough solar: feed 1MW continuously via 23.8 panels, 21:25 accumulator ratio. Block-power planning prevents the "I grew but production died" classic loss. Separately metering each module's power adds management but kills surprises.

Beacon – Factorio Wiki

wiki.factorio.com

Station Design: Zero Main-Line Waits

Flip to trains? Never park a train on the main line. Scale up, pain shifts from "can trains move?" to "where do stopped trains choke?" Shared stations without stackers grid instantly. One train waits for a dock, the next blocks the junction, that propagates, and the whole network locks.

I lived this: unsacked multi-train main line, one train parks for cargo, next queue forms, opposite-side trains also lock, entire supply choked. Slap a stacker in front? Waiting queues behind the station, main line keeps flowing. Science output visibly bounced. Main-line-wait-free is the mantra.

Operationally: stations receive; trunks flow. Stations get multi-lane wait zones; arrival trains slot in order. Same resource station feeds multiple ships without mainline collapse. Tutorials: Train Signals detail this, but the pre-req is parking outside the main road. Congestion isn't train shortage—it's no-wait-space design.

This meshes modular design: each block owns "input dock" + "factory" + "output dock," aborbing wait-queues within the block. Main lines stay high-speed commutes. City-block geometry standardizes station placement, growing stations like modules. Late-game train-base's real work isn't laying rail—it's stamping out "station + stacks" in volume.

Space Age Reshuffles the Design Rulebook

Inter-Planetary Spread and New Transport Logic

Space Age kills the "build one mega-main-bus on the ground" assumption. Released 21 October 2024, the DLC adds 4 new worlds, space platforms, re-shuffled tech trees. The engine shifts from "optimize one ground map" to "optimize per-planet, then connect platforms."

Resource terms break across planets. Ground main buses work because rich central iron/copper supports one vein model. Space Age breaks it: Planet A's copper abundant, Planet B's scarce. Same design breaks both. I tried copying ground patterns, hit shortages instantly, then realized: modular per-planet + selective export beats planet-spanning mega-bus.

This ripples into logistics. Ground factories can "dump to trunk"; planets can't. Export costs matter more; waste stings faster. Doctrine flips: make mid-tier locally, export only value-dense product across planets. Copper plate hauls cheap; copper-as-circuits costs bandwidth. Platforms become selective hubs, not universal vacuum cleaners.

Main bus isn't dead. Each planet's one site? Still valid, local scale. Role shifts: local optimized spine instead of universal architecture. Each planet: isolation, strip-mine, pre-process, export carefully. Cross-planet supply doesn't snake through one mega-trunk; it hub-and-spoke through platforms.

Space Age

wiki.factorio.com

Quality: The Vertical Dimension

Space Age's Quality feature rewires optimization thinking. Four tiers now exist for every item. Old factories asked "how much?"; now it's "how much, and what tier?"

Historical design: same-quality volume flow. Quality adds: pick-out high-tier pieces, route carefully. Dumping gold-tier circuits into generic lanes wastes value. Selectivity becomes design content.

Main bus + quality = tension. Trunks flow unqualified mix, but gold-tier items demand segregation. No branch-point logic handles "quality-aware split." So cleaner: dedicated tiers, staggered routing. Gold circuits to premium gear only; standard to mass production.

Scale-wise, it changes the math. Can't just "add a 5th circuit line." Now track: how many gold, standard per destination? Dimensionality rises. Main-bus designs feel cramped because trunk and quality-tiers collide. Result: Space Age favours quality-aware forks—more complex control logic, more floor space, but value optimization kicks in.

Practical shift: every mid-tier gets quality segregation. Make locally, filter by tier at consumption. Space Age ends up looking less like "wide bus" and more like "filtered module sets." Design evolves from throughput-only to throughput-and-grade.

Fail Patterns & Repairs

Wreck Checklist by Type

Collapses usually look different but trace to few roots. I spot belt ceiling amnesia and no-growth buffer space most.

Cramming too much into yellow trunks: 15 items/sec = hard stop. Iron thin downstream? Shift wire-making to circuit zones, not from a shared trunk. Demand stays constant; belt saturates fast. Don't scale belt; reroute consumption. Smelting + first-layer processing + export: that chop bleeds load.

No future space: Main bus thrives if you pre-reserve "room for growth." Adding iron/copper/circuit lanes kills it if you gridlock lanes early. Tag 4-lane voids at startup "future circuits," "future steel," and don't filch them. Orthogonal building after a point destroys signal. I tag zones left/right/up/down and lock approach, not grab what feels empty.

Stackless rail stations: Multi-train, no stacker = main line parking lot = instant gridlock. Symptom: one train waits dock, next queues junction, opposite trains lock, everything stops. Cure: station = platform + multi-lane stacker. Waits live inside the station envelope, main line flow—untouched. Rails need discipline upfront; retrofit is rework-heavy.

Early beacon-rushing: Beacon ×N = 480kW×N. Cramming density before flow-logic settles = high stats, zero output. UPS tanks, production freezes. Real sequence: **sort logistics → split blocks → then beacon-ify**. Density without flow is a power sink.

Power after-thought: Grid brownout = entire factory freeze, not just one belt. Rig budgeting upfront: solar ×23.8 per 1MW, accumulators at 21:25 ratio, nuke for step-loads. Adding blocks without new powergen kills everything, not the block.

Space Age neglect: each planet self-isolates? Nope—locals defend, plants cross-feed, decay eats exports. Segregate defense, pre-designate recovery order, throttle cross-planet trades. You collapse if one supplier goes rogue; planet-level resilience is design content now.

Making Factories Recoverable

Unbreakable beats repaireable eventually. Unbreakable factories fail anyway, so build recoverable instead: that pays dividends long-term.

First: compartmentalize by role. Smelting, circuits, science, oil, power—distinct zones, clear IO. Shortage = quick source (local block short) or (trunk can't deliver).