Most Sandustry transport failures begin before the first belt is placed: the player has chosen a route without first classifying the material. Solids belong on Conveyors and Launchers. Ordinary liquids use a Pump, Pipes, and a Liquid Vent. Molten material is excluded from that liquid network. Filters separate resources, but the Basic and Advanced versions do not accept the same states.
This guide uses the August 13, 2026 Early Access launch rules. It does not assign invented pressure limits, pipe lengths, pump ranges, throughput percentages, or power costs. The launch guide states that the logistical structures covered here require no power and continue operating regardless of whether material is present. If the client changes during Early Access, use current placement feedback and patch notes.
Classify the cargo before choosing the route
| Cargo state | Supported transport | Supported separation | Important limit |
|---|---|---|---|
| Solid | Horizontal Conveyor; vertical or diagonal Launcher | Basic or Advanced Filter | Each Filter selects one resource |
| Ordinary liquid | Pump to Pipes to Liquid Vent | Advanced Filter | A Vent cannot choose among mixed liquids |
| Gas | Use the current researched process shown by the client | Advanced Filter | The launch guide verifies filtering, not a universal gas transport layout |
| Molten fluid | Convert state before moving | Do not assume an ordinary liquid line applies | The normal Pump/Pipe/Vent route explicitly excludes molten fluids |
The table is a decision gate, not a recipe list. If a resource will not enter a Belt, inspect it with the Material Scanner before adding more Belt pieces. If a liquid will not behave in an ordinary pipe route, ask whether it is molten or whether the Pump and Vent level conditions are satisfied. A device mismatch cannot be fixed by making the route longer.
Conveyor direction is decided while you drag
Conveyor Belts move solid resources horizontally. Their direction follows the direction in which you drag during construction. Build the receiving side first or leave a clear marker at the destination, then drag the Belt in the intended travel direction. Feed a small, distinctive batch before connecting a full storage pile.
A working Belt test has three visible states:
- The solid lands on the active surface instead of beside or under it.
- The Belt carries the material toward the destination.
- The material leaves the Belt where the next structure can actually accept it.
If the first condition fails, move the input or add a controlled drop. If the second fails, rebuild the reversed segment; there is no reason to redesign the entire factory. If the third fails, inspect the receiving structure’s active area and collection space. A correct direction can still produce a pile against a wall when the exit is misaligned.
The early Gold line is a useful test module. Wet Sand arrives at a Shaker, Gold drops into its collector, and residue leaves on a separate Conveyor. It makes direction mistakes obvious because the two outputs have different destinations. Do not route both into one anonymous pile and hope to separate them later.
A Launcher always throws from its highest point
Launchers move solids vertically or diagonally. The highest point of the Launcher shaft is the output, while the drag direction determines whether the material is thrown left or right. Read that sentence as two separate configuration checks: height chooses the exit; drag direction chooses its side.
When a Launcher appears idle, inspect it in this order:
- Intake: is the solid entering the low end of the shaft, or piling one tile away?
- Shape: does the Launcher actually rise toward the intended floor?
- Throw side: is the highest point ejecting toward the next structure rather than away from it?
- Landing: is there an open receiving area, or does the material hit a wall and fall back?
Use the shortest Launcher that proves the handoff, then extend or copy it. A very tall shaft hides whether the fault is at the intake, output side, or landing. Mark 2 Launchers increase velocity after Research unlocks them, but more velocity does not change which end is the output.
Allow and Block solve different sorting problems
A Basic Filter works with solids and selects one resource. Its two modes reverse what passes through:
- Allow: the selected resource passes; everything else continues along the other route.
- Block: the selected resource is held out of the pass-through route; other resources may pass.
Use Allow when one valuable or dangerous material must be captured immediately. The launch guide’s simple example is an Allow Gold Filter: Gold takes the selected exit while the rest continues. Use Block when the mixed stream contains two known materials and peeling one away makes the other easy to collect.
The one-resource rule matters. A Basic Filter cannot Allow both Redsand and Moonhop as a combined selection. One supported arrangement is to Block Moonhop at the first Filter so Redsand can pass, then Block Redsand at a later Filter to isolate the second material. The important principle is sequential classification: each Filter answers one yes-or-no question.
Advanced Filters extend the same Allow/Block and one-resource logic to Solid, Liquid, and Gas resources. “Advanced” does not mean one block can select an arbitrary list. If a mixed line needs three outputs, plan several named decisions and leave a reject path. A Filter without a safe destination for the non-selected stream simply moves the jam to another side.
Ordinary liquids use Pump, Pipe, and Liquid Vent as one system
The supported liquid route has three parts:
- The Pump is the input.
- Pipes connect the source and destination.
- The Liquid Vent is the output.
Pumping continues while fluid exists at the Pump’s level or above it and the destination fluid has not risen to the Liquid Vent’s level. Those conditions produce two very different-looking stoppages. A source that falls below the Pump starves the line. A destination that fills to the Vent closes the available output. Before rebuilding any Pipes, look at both ends.
A single pipeline can contain multiple ordinary liquid types, but the Liquid Vent cannot be told which one to release. It prioritizes the closest liquid. That makes a shared line physically possible and operationally unreliable for a process that requires a specific input. The safe launch-build practice is to separate liquid lines by purpose.
If the wrong liquid appears, do not add a Basic Filter—it handles solids only. Do not assume that a longer pipe will create order. Empty or isolate the mixed line, connect one source, and observe the Vent. Use an Advanced Filter only when the required Research is unlocked and a deliberate one-resource separation is part of the design.
Molten material needs a state-change handoff
The normal liquid network excludes molten fluids. This is the boundary that prevents a common wasted build: placing Pumps and Pipes beside lava as though it were ordinary water. The launch guide provides a different relocation route. Freeze lava into solid Scoria, mine it so the loose resource becomes Cinder, move that solid with ordinary solid logistics, then heat the Cinder at the destination to make lava again.
From a logistics perspective, this is four separate modules:
- a controlled freezing area that produces Scoria;
- a mining and collection step that yields movable Cinder;
- a solid Cinder transport route;
- a contained heating destination where lava may safely re-form.
Test each module with a small amount. If Scoria exists but no Cinder reaches the Belt, the mining or collection step failed. If Cinder reaches the destination but lava does not appear, the transport worked and the heating step did not. Keeping the modules distinct prevents the entire build from being blamed on “bad logistics.”
Copy modules only after the reject path is visible
Select an existing structure with C; use Ctrl+C to copy it and Ctrl+V to place another. The default Ctrl replace control exchanges an existing structure for a later tier. These shortcuts are ideal for repeated Filter cells, Launcher lifts, and collector modules, but only after both the selected and rejected outputs have completed a test.
When Mark 2 Conveyors or Launchers unlock, replace transport capacity where a proven route is actually saturated. Upgrade signals include a continuous input pile while the receiving machine stays hungry, or a Launcher that completes every throw in the correct direction but cannot clear material as fast as it arrives. A wrong Filter mode, a full collector, or a liquid Vent covered by its destination pool is a logic problem, not a speed problem.
Run this diagnosis from the first missing material
- Identify the expected material and its current state.
- Find the first checkpoint where it fails to appear.
- For a Conveyor, check landing, drag direction, and exit alignment.
- For a Launcher, check low-end intake, highest-point output, throw side, and landing.
- For a Filter, check Basic versus Advanced, Allow versus Block, and the one-resource selection.
- For a liquid line, check source level at the Pump, destination level at the Vent, and unwanted liquids in the Pipe.
- For molten material, stop testing ordinary liquid logistics and move it through a supported solid-state conversion.
Once one small batch reaches the correct destination and every rejected output has somewhere to go, reconnect continuous input. At that point the route is a module worth copying, not an experiment hidden inside a factory wall.
Sources & credits
The official Steam page and release announcement set the Early Access identity and resource-simulation boundary. Device behavior for Conveyors, Launchers, Filters, Pumps, Pipes, Liquid Vents, and molten materials comes from the developer-linked Sandustry Beginner’s Guide. A launch-window hands-on guide supplied a second check on the opening Shaker output layout.