If every belt looks busy but the target output barely moves, stop adding Miners. In Modulus: Factory Automation, the visible slowdown is often upstream of the real restriction: a full output, an undersized consumer, a stale Freight Hub dock, or an operation performed after the block was divided too early.
This guide uses Happy Volcano’s official Academy examples and is current through Patch 1.1.1 on July 6, 2026. It assumes the relevant operators are already unlocked. It does not prescribe a permanent universal ratio; the target module, Tech Tree upgrades, travel distance, and live operator panels decide the correct capacity.
Begin at the consumer and walk against the flow
Open the building, crane, Data Center, or Delivery Dock that is missing output. Record what it wants per minute, then inspect the last operator that should supply it. Moving backward prevents a common mistake: upgrading the extraction end while the final Assembler or crane is already saturated.
Unlock Efficiency View in the Tech Tree and press H. Green means an operator is running at full capacity, yellow marks underperformance, and red signals a serious restriction. The view also uses patterns and numbers, so its diagnosis is not dependent on distinguishing red from green.
| What you see | Test next | Likely response |
|---|---|---|
| Empty input and idle output | Follow the missing ingredient upstream | Repair supply, routing, or the recipe before adding consumers |
| Full input and empty output | Open the operator configuration and target preview | Correct orientation, recipe, production resource, or operator state |
| Full output belt | Inspect the next machine, crane, hub, or filled requirement | Add consumption or redirect valid overflow |
| Green machine behind a stalled line | Check whether its output is backing up | Look farther downstream; green does not prove the whole branch is healthy |
| Entire factory halts at once | Confirm version, mods, and a clean reload | Rule out the fixed client issues before rebuilding |
Use counters where a visual belt is too dense to read. Compare the rate before and after the suspected operator. A rate drop locates the constraint; matching rates with a full belt indicate that the consumer, not that section, is setting the pace.
Bisect a blocked line with a temporary Scrapper
Happy Volcano’s debugging method uses a Scrapper as a controlled pressure release. Place it after the area you suspect. If the upstream operators recover in Efficiency View, the problem is downstream of the test point. Remove it, move closer to the consumer, and repeat until the transition is isolated.
This is faster than deleting a long belt because it changes one condition: downstream capacity. In the official example, two Cutters together produced 30 items per minute into a consumer that accepted only 15. The Cutters were not broken; half of their combined output had nowhere to go. Splitting the flow across a second downstream operation restored the intended rate.
Do not leave the diagnostic Scrapper in place without checking what it destroys. It is a probe, not automatic overflow design. Once the constraint is known, replace the test with the actual fix: another consumer, a corrected branch, a slower upstream cell, or an intentional surplus route.
Let splitters allocate by demand
A normal Splitter does not insist on a permanent 50/50 division. When one branch is unable to accept more items, additional flow moves to the open branch. That behavior supports a compact distribution spine: run the source past a row of identical consumers and peel off material at each stop.
The official Dye Mixer example starts with a 30-per-minute supply and four Mixers that each consume 7.5 per minute from the relevant resource. A sequential row feeds all four without building a balanced tree in advance. Each Mixer takes what it can use, while the remainder continues to the next.
- Keep the two ingredient spines visually distinct so an Oil problem cannot be confused with a pigment problem.
- Collect outputs in a tunnel or separate return belt instead of crossing them back through the input row.
- Leave an inspection gap before the last consumer; if material reaches that point while an earlier machine starves, the branch is misconfigured.
- Duplicate the whole tested cell only after the last consumer holds its intended rate.
This demand-driven pattern is best for identical consumers. A module line with several shapes or colors still needs explicit separation. Convenience is not worth allowing the wrong intermediate to reach an Assembler input.
Apply paint before geometry multiplies
Paint is charged per object processed, not per voxel inside that object. The costly order is to cut a Cube into many pieces and paint every piece. Reverse those steps when all descendants need the same color: paint the largest shared object, then cut it.
In the official Android Assembly lesson, three separate Painters consumed 45 paint per minute for 45 modules. Moving the painting step to the raw Cube kept the same module output while reducing paint demand to 15 per minute. The geometry did not change; only the order did.
There is one guardrail: do not pre-paint a shared block when later descendants need different colors. Find the latest stage at which every voxel in the object still has the same required color. Paint there, split afterward, and keep divergent colors on separate branches.
Patch 5 also makes this logic easier to route through Stampers. You can choose which stamped result exits each side, and you can decide whether a blocked scrap side should stop both outputs. Check those toggles before twisting belts around a layout inherited from the launch build.
Process several future modules as one object
The same per-object principle applies beyond painting. Instead of constructing one module, repeating an operation, and copying the entire line, arrange several future modules inside a larger block. Rotate them so identical cuts, colors, or voids align. Perform the shared operations once, then make the final cut that separates the finished modules.
Happy Volcano calls this the sushi roll approach. In its Lumite Reactor example, the spacious layout took 240 Cubes per minute and produced 240 modules per minute. The reorganized layout fed 60 large blocks per minute, still produced 240 modules, reduced the number of repeated operators, and used one quarter of the paint.
The important step happens before any building is placed:
- Duplicate the target module mentally or in the module workspace.
- Rotate copies until regions with the same treatment form continuous volumes.
- Apply Monotoner, Painter, Stamper, or Cutter operations while those volumes remain joined.
- Separate the repeated modules only after the last shared operation.
- Wrap the proven chain into a rectangular, copyable production cell.
Disconnected pieces can also form an intermediate. Assemble several floating parts into one cluster, process that cluster once, and attach it to the main body later. The operator does not require every voxel in an intermediate to be physically joined. This can turn four identical color or stamping operations into one.
Balance buildings in multiples of the real consumer rate
The launch Academy examples revolve around 15-per-minute steps, but the correct count comes from the live panels. One early line has four Miners supplying 60 Polyrock per minute to a Furnace that creates 15 Cubes. A Cutter can turn those Cubes into 60 smaller pieces, while an Assembler consumes 15 pieces per minute on each of two inputs. One Assembler therefore leaves half that Cutter output unused; two parallel Assemblers consume the complete 60-piece stream.
Order can remove machines as well. Happy Volcano’s alternative first assembles larger intermediates and cuts afterward, reaching the same final piece rate in less space. Compare both designs by four numbers: raw input, operations per minute, final output, and occupied footprint. A layout is not better merely because every belt moves.
Buildings add a ratio layer. The official Basic Blue Pigment Refinery example needs one construction module 32 times and two others 16 times each, a 2:1:1 relationship. Producing those at 30, 15, and 15 per minute matches cranes that deliver 15 per minute each: two cranes for the first module and one for each other module. After construction, check the separate production resource. Perfect construction ratios cannot operate a refinery that has no Hexacrystal supply.
Assign Freighter routes by dock number
A Freight Hub has four numbered docks, and each number maps to the matching cargo slot on the Freighter. Material entering dock 1 fills slot 1. Slots 2 through 4 remain empty unless their corresponding docks receive something. Naming a hub Blue Paint helps humans, but it does not override that slot mapping.
Open Freighter Control from the top menu or the hub, name and color the route, then define each stop. Under the documented control scheme, click a slot once for loading and again for unloading. Apply the route and watch one full circuit before trusting it.
If the resource connected to a dock changes, clear the old contents. The dock can keep expecting its earlier item, producing a route that looks configured but refuses the new cargo. This is the first check when a newly repurposed hub stops moving goods.
Freighters can distribute overflow without an exact split. One official example carries 240 blue paint per minute to two destinations consuming 120 each. The first hub fills, retained cargo continues to the second, and the loop settles without a separate 50/50 route. Distance still matters: if the round trip is slower than consumption, add a second Freighter on the same route or shorten the circuit.
A return leg can carry another resource in a different slot. Write the slot plan into the hub names or colors before using a multipurpose route. Saving one vehicle is not an optimization if an accidental cargo swap stops two islands.
Upgrade coolant before multiplying advanced operators
Advanced operators need Coolant through Supply Tanks. Patch 5 sets base efficiency at 16 operations for each unit of Coolant, then offers upgrades to 32 and 64. Happy Volcano’s reference case puts an advanced operator at 120 operations per minute, which consumes 7.5 Coolant per minute before those upgrades.
Use that boundary to choose between another Chemical Plant and a Tech Tree upgrade. If several advanced operators are about to be copied, the efficiency node reduces every connected demand. If only one remote cell is short, repair its Supply Tank, drone link, or route before expanding the global coolant system.
Patch 5 also doubled the Freighter cap. Do not interpret the higher ceiling as a requirement to replace every Skyline. Move high-volume, shared resources by Freighter where it simplifies several islands; keep short, stable local links when the route would add delay and dock bookkeeping.
Rule out version and mod state before redesigning a stopped factory
Patch 1.1.1 fixed two whole-factory stop causes: thread-safety trouble in Statistics/Achievements and a problem connected to the Salvager. It also fixed Freighter jitter, Harvester Pad exceptions after moving with undo/redo, and lost non-shape progress in GNN buildings. Earlier updates addressed random Cutter stops, multiple-crane behavior, drone links, long conveyors, and mirrored Assembler configurations.
If a previously stable factory freezes everywhere, save, confirm Patch 1.1.1 or newer, reload, and test the same area without mods. The June mod-support update states that achievements do not unlock while any mod is active and that a save used with mods remains permanently marked as modded. Disable mods to reproduce a fault, but do not expect that step to restore an achievement-eligible flag to that save.
A production cell is ready to copy when its consumer holds the required rate, its surplus has an intentional destination, its Efficiency View state is understood, and its dependencies survive a save/reload. Copying before those checks scales the uncertainty along with the output.
Sources & credits
Valve’s official Steam listing identifies AppID 2779120, Happy Volcano as developer, Kwalee as publisher, and the April 2, 2026 release. The official game site defines the cut-paint-stamp-assemble system and its no-timer sandbox character.
The debugging, ratio, paint, block-processing, Freighter, coolant, mod, and patch rules come from Happy Volcano’s official Steam announcement feed through Patch 1.1.1. Two readable Community guides carry removed/incompatible warnings and were used only as limited cross-checks for large-block production; a conflicting fan SEO wiki, trainer advertisements, an unattributed cheat page, and a joke guide supplied no optimization facts.