Transport Use Cases

2026-05-10

Comparison and guidance on when to use belts, trains, and logistic robots in Factorio — translated tutorial guide.

** This page compares the three main transport methods—belt transport systems (conveyor belts), trains, and logistic robots. Also useful as means of transport are the character's inventory and vehicles. (Vehicles in particular have very large carry capacity and can sometimes substitute for trains, although you must drive them yourself. See the separate vehicle chapter for details.)

The optimal choice depends on the situation: for example the amount of resources to move and the distance.

Comparison: belts, trains, robots

| Belt | Train | Logistic robots |

Throughput | Steady, low per line, exactly calculable, limited | Very high, fast, hard to calculate exactly but stable; effectively unlimited if you have space | High in small areas, unpredictable/chaotic, unlimited with enough bots, not optimizable, deadly for long-distance/mass transport |

Scalability | Low. You cannot reuse existing belt lines; you must lay a new belt from A to B. | Medium. Stations A and B can connect to any point on an existing rail line. You don't need separate lines; trains run on existing tracks. Improve fuel or lengthen trains to raise throughput without laying new lines. | High. Chests A and B can be placed anywhere on the network, including existing chests. |

Flexibility | High. Small and with underground belts, you can ship almost anywhere except highly dense built areas. Excellent. | Medium. On trunk lines you can move any item from anywhere to anywhere. But stations, junctions and turnarounds require lots of space. Very limited room to lay tracks without significantly modifying existing buildings. | Almost infinite. Delivery needs as little as 2 tiles (chest + inserter). No restriction on item variety; a single chest can request many kinds. |

Required space | Small for simple products, large for complex product chains | Large for stations and corners. There's a difference between RoRo-style (large) stations and terminus stations (quite small). | Small even for complex products. |

Optimization | Highly optimizable. Endless fun. | Often not necessary—it's better to rebuild or add tracks. Large networks can suffer deadlocks/congestion. Use chain signals to optimize. Optimize station connections: connect by belt (complex) or by logistic bots (recommended). | Optimize placement of roboports (for charging) and the ratio of ports to the number of flying bots. |

Parallelization | High. Two parallel belts are twice the throughput of a single belt. Cost is roughly double, and construction effort doubles. | Extremely good. With one-way-running trains, two parallel lines can carry 5–10× more trains than double-headed lines because blocked traffic can choose the other line. Laying two parallel lines often costs less than double the effort. | Not great. Increasing bots from 1 to 2 doubles throughput, but increasing from 500 to 1000 can cause complete chaos. |

Initial cost | Very small if belts are short. Parallel belts are still good. Cost rises when making fast belts. Express belts are expensive and should be used only in special cases. | Early on, locomotives require significant materials. Mid/late game costs are trivial. Rails are cheap relative to increases in throughput. | Roboports and especially bots require lots of resources. Upgrading logistic bots is very expensive and should be accounted for in total cost. Maintenance is time-consuming. |

Energy use | Free. Always debated. ;) | Low. Fuel use is currently cheap. | High. Roboports consume large amounts of electricity, and power use grows with the number of flying bots. It can be a big issue, though usually manageable. |

Items on transport vs transport time | Bad. A 100-tile belt holds about 700 items when full, but a basic belt takes 56 seconds to move that length. Even an Express belt is only about 3× faster. | Very good. Can move large quantities in a short time. | Chaotic. When clogged, the most important flying items can become inaccessible. |

Maintenance cost | None | Some fuel required | Considerable electricity required |

Pollution | None | Yes | None (though indirect effects from electricity use) |

Best uses | High throughput, bulk materials (ores and some intermediates), short–medium distances. Ideal for raw materials and simple products. | High throughput, long distances, single-cargo runs. Example: moving ore or plates from a deposit to a main factory. Supplying outposts (train + drones, repair packs, diesel, ammo). | Extremely high throughput at very short range (under 50 tiles); low–medium throughput at midrange (50–500 tiles); rare for long distances. Ideal for products that need many types in small quantities (mall buildings) or for refueling locomotives if you don't refuel at a separate station. Unbeatable for station unload/load chests. |

Biggest headaches | Left/right lane problems (game mechanics), adjusting splitters for proper distribution. | Deadlocks at complex intersections, complicated station and signal setups, congestion (especially when a locomotive gets bit and partially blocks track with its wagons). | Roboports blocked waiting for charge, "dumb" behavior like distant bots being used while nearby bots are idle, bots taking "shortcuts" through congested areas. |

UPS cost per item (relevant only on weak hardware or when science/minute reaches five digits) | Variable. Long partially filled lines or many inserters and balancers increase cost. Short fully compressed lines with minimal balancers and few inserters cost little. | Low. | High. |

Expert notes Belts

Practical distance is about 500 tiles (for comparison: radar has a max range of 200 tiles, covering radius 100 tiles).
Belts are useful for connecting deposits to factory areas. A fully compressed yellow (basic) belt can carry up to 900 items/min. Using fast belts or express belts increases this (up to 2700) but at huge cost. Parallel belts increase cost but basic belts remain economical even for long distances.
Best for bulk materials within automated facilities. Examples: ores, plates, electronic circuits for short–medium transport.
For long distances, belts provide a hidden buffering amount (1 tile holds 8 items per lane, so two lanes over 500 tiles = 4000 items!). That buffer can become unreachable and is effectively wasted until transported.

Note: the 500-tile rule of thumb is based on:

A 500-tile belt takes about 5 minutes (4.44 minutes) to travel end-to-end.
A 500-tile belt stores 4000 items (equal to one cargo wagon's worth).
Beyond roughly this length, trains become more reasonable.

Belt advantages:

Belts always provide a constant throughput.
Simple calculations like "do I need more belts to carry materials from the mine?" are easy.
Basic belts are cheaper and easier to build than tracks.
You can visually see how many items are on a belt and easily judge if throughput is sufficient.
Belts can be used as open storage: two lanes across 10 belt-lengths store 80 items (though this is a drawback for long belts).

Disadvantages:

Throughput is limited.
Slow. Very long belts take a long time to fill; items on the belt are unusable until they arrive. A belt of length 100 holds 714 items for 56 seconds. Trains reduce this wait to about 10 seconds, and items on belts are considered "unavailable storage" until they arrive.
Items on belts are "unusable storage" until transported.

Belts are continuous and provide the highest throughput for direct short-distance factory connections. Especially when using parallel basic belts, you get very high throughput at low cost.

Trains

Use trains for large, distant resource sites. Nothing is faster over long distances. A single rail line is easy to lay, but planning a large rail network takes time and skill. A clever trick is to ride a locomotive and use the rail planner to lay tracks ahead of you.
Trains are also good for connecting remote factories, allowing you to "ride to" an outpost to service it.
Smelting ore on-site into plates and shipping plates by train doubles per-wagon efficiency and lets you send materials directly to storage yards or factories.

Logistic robots

Robots are best for limited zones with high building density—typically the central factory area. Updates have made robots capable of handling huge numbers of items.
Robots are not a good fit for long distances or bulk materials: using bots for ore or furnace output is not advisable. It can be done, but AI limitations often make it inefficient; don't expect it to work perfectly. Some players enjoy this as a challenge.
At train stations, if robots can charge between jobs they can handle large throughput perfectly: train arrives and unloads, bots carry to target chests, bots charge. Next train arrives.
Robots are ideal for transporting expensive items sporadically or for delivering intermediate goods inside a local network where laying belts isn't practical.

Robots excel at moving slow-produced, rare items but are extremely costly for inclusion in mining/smelting mass transport due to frequent charging breaks.

Exceptionally, robots are very effective around train stations. It's often hard to lay belts around trains to properly distribute load due to space and non-continuous flow, so robots benefit from the charging downtime and work well.

From a throughput perspective

A basic (yellow) belt carries up to 900 items/min, and an express belt up to 2700 items/min (see above). Compare that to trains: a cargo wagon carries 2000 items (ore) per wagon! If you refine on-site into plates (iron/copper), a wagon can carry 4000 items. So in throughput terms you need to switch to trains or lay multiple parallel belts.

You might think logistic robot research changes this, but it doesn't. Robots are strongest in relatively small areas with research bonuses and try to spread deliveries evenly. But robots are weak for long distances and bulk. The following calculations illustrate this (assuming fully upgraded logistic bots):

Number of robots required to move 3000 items/min over 50 tiles:

Robot speed: 3 tiles/sec (base) × 1.4 (logistic speed bonus) = 4.2 tiles/sec Round trip time: 100 tiles ÷ 4.2 tiles/sec = 24 sec Bots needed per 1 item/min: 24 sec ÷ 60 sec = 0.4 bots Considering bot carry capacity: 0.4 bots ÷ 4 items/transport = 0.1 bots (for 1 item/min) Total: 3000 items × 0.1 bots/item = 300 bots

300 is high but realistic for a logistic network with 4–5 roboports.

Number of robots required to move 3000 items/min over 500 tiles:

Round trip time: 1000 tiles ÷ 4.2 tiles/sec = 240 sec Bots needed per 1 item/min: 240 sec ÷ 60 sec = 4 bots Considering bot carry capacity: 4 bots ÷ 4 items/transport = 1 bot (for 1 item/min) Total: 3000 items × 1 bot/item = 3000 bots

3000 bots take considerable resources to produce. This ignores research (bot speed/stack bonuses), number of roboports, and required power, which also add heavy burdens.

Then your choices narrow to belts or trains.

Trains are faster than belts and offer better logistic control when moving between logistic networks.

Belts simply take a long time and tie up many assets as "in transit." Items on belts should be considered "unavailable storage." Backed-up belts are useful as buffers, but long belts are inefficient: a long gap can hide that the far side is depleted for up to 5 minutes.

You can squeeze the last 20% of belt performance depending on distance and total volume, but at some point trains win. In the central factory area robots remain unbeatable. For simple single-item flows (e.g., iron ore → iron plates), parallel belts give cheap, efficient huge throughput. In the late game fast belts are a cost-effective compromise; express belts should be used only on short runs.

Required bot counts depend directly on building density. The factory core should be dense because both throughput and speed (including transport time—how fast one item is produced overall) matter.