Sandustry automation

When to automate

Automate after Shaker exists, Wet Sand is stable, and walking the same trench is slower than digging a new one. Automate before you place a bank of Shakers you cannot feed. A starved Shaker is a research delay dressed up as a factory.

• One material per trunk until Filters exist.
• Feed Shakers from the high side of a diagonal.
• Launchers need a receiving bin. Overshoot is lava or void.
• Debug each phase before you stack them vertically.

Three-phase refinery

Phase A is the gold core: sand dig to belt to wetting pool to Wet Sand belt to diagonal Shaker to Gold pit. Phase B is the Residue tail: Shaker output to burn lane to Burnt Residue to Kinetic Press to Gold merge. Phase C is the vertical link: launcher stairs from a deeper sand cut back to the surface deck.

Each phase should run alone. A monolithic tower is un-debugable when steam, slag, and seeds share one shaft. Use Blocks to build hoppers. Use Collectors on the Gold pit. Use Lights before you add Digger drones underground.

Filters and drones

Filters are insurance, not architecture. Deploy them when a rebuild mixed Sand into Wet Sand, when Residue walks back into the pool, or when Burnt Residue must stay off the Press input. Community density talk places Lava between Redsand and Gold. Use that only as a sorting hint, not as a fake stat block.

Digger drones keep sand income up while you research. Hauler drones bridge cliffs you have not belted. They are worse than belts at defining geometry and better than you at repeating a haul. Buy Logistics first. Lights help them path. Demo Hotfix 7 hover-through-buildings is for you, not for the drones.

Reading a jam

Empty Shaker input means the Wet Sand belt or the wetting pool died. Look at Water first, then dry Sand. A belt that backs up at a launcher needs another stage or a wider bin, not a faster belt you do not have. Gold down and Residue up means the burn lane is slower than the Shaker. A Kinetic Press sitting idle means Burnt Residue is not arriving, which usually means the fire went out or lava froze into Scoria in the trench.

Random wrong pixels on a Gold belt are a Filter problem or a missing wall. Steam bubbles in a sand trunk are a heat problem. Seeds on a Residue belt are actually useful if you planned a farm, and a disaster if they sprout on the burn strip. Walk the line once after every research buy. Hover through buildings if you must, then still cut a corridor so the next jam is visible from the tank.

Mk.2 belts exist so long spines do not become the bottleneck. They do not fix a bad junction. Short trunks, one product, labeled in your head as sand-in, wet-out, residue-out, gold-home. That vocabulary is the whole automation skill.

More detail worth keeping

Think in pixel physics, not in Factorio muscle memory. A belt here is a moving ledge under grains that still have density and wetness. If you treat Wet Sand like a stackable item, it will slump, mix, and lie to your Filter. Keep trunks short enough that you can see the product. Use Launchers as stairs, not as artillery. Use drones as temporary bridges, not as a religion. Lights are cheap compared to a lost Hauler in a black shaft. When you copy a screenshot from Steam, copy the clearance above the deck, not the pretty curve of the belt. Hover-through-buildings is a debug cheat the community voted in. It is not an excuse to omit walkways in a factory you intend to keep after the next hotfix.

Practice that survives a patch

A launcher stair is three tiles of honesty. The bottom needs a bin, the middle needs clearance, the top needs a belt that is already moving. If any of those is missing you have a cannon, not a lift. Build the bin first. Then the launcher. Then the belt. Reverse that order and you will spend a night vacuuming Wet Sand out of the sky. Filters belong after that stair exists, because the mix you are afraid of has not happened yet.

Close the loop

Name belts by product in your head every time you come back from a vault. Sand in. Wet out. Residue down. Gold home. If you cannot say the four names while hovering the deck, the factory is already mixed. Rename by rebuilding the junction, not by adding a Filter as a confession.

A pipe crossing a belt should be visually boring. If the screenshot looks clever, it will leak. Run fluids behind the deck in a Block trench. Run solids on the deck. That split is the entire mid-game aesthetic of a factory that still runs after a hotfix.

Frequently asked questions