Satisfactory Trains: Power, Signals and Station Throughput

Trains are the only way to move large volumes across the map without a belt running the whole distance. They are also the system most likely to brown out a grid that was perfectly stable before you built them.

Power: the number that catches people

An Electric Locomotive draws between 25 and 110 MW. Both ends of that range matter:

• 25 MW is the floor. A locomotive draws it constantly, even parked and idle. A train sitting at a station is still costing you power.
• 110 MW is the ceiling, hit while accelerating a loaded train, especially uphill.

Size your grid for the peak, not the average. A four-train network idling costs 100 MW; the same network with all four accelerating at once briefly wants 440 MW. If your generators cannot cover that spike, the grid trips, the trains stop, and now nothing is moving.

Stations add their own fixed cost:

Building	Power
Train Station	50 MW, constant, docked or not
Freight Platform	50 MW each
Fluid Freight Platform	50 MW each

A station with two freight platforms is 150 MW before a single train arrives. This is usually the bigger surprise: people budget for the locomotive and forget the infrastructure is drawing more than the train.

Speed and the downhill problem

Top speed under power is about 120 km/h on flat track, less when climbing.

Downhill is a different story. A train rolling down a long slope keeps accelerating well past its rated speed — the wiki puts it near 500 km/h given enough drop. Braking force is 600 kN above 80 km/h and 2000 kN below it, so a train that has run away downhill has the least braking authority exactly when it needs the most.

The practical rule: keep gradients gentle, and if you must descend a long way, do it in stages with flat sections between rather than one continuous drop.

Block signals versus path signals

Both cost the same (2 Steel Pipe, 1 Computer). They solve different problems and using the wrong one is the usual cause of a deadlocked network.

Block Signal divides track into blocks. One train per block. Use these on plain line, between stations, anywhere track is simply going somewhere.

Path Signal reserves a route through an intersection rather than the whole area. Use these at junctions, crossings, and anywhere two routes overlap.

The failure mode that traps people: putting block signals around a junction. Two trains each claim part of the intersection, neither can proceed, and the network stops permanently. Path signals exist precisely to prevent that, because a train will not enter until it can reserve a complete path out the other side.

Simple rule that avoids nearly all deadlocks: block signals on open track, path signals at every junction.

Station throughput

A Freight Car holds 32 slots. What that means in items depends entirely on stack size, which is why the same train moves vastly different quantities depending on cargo.

The real ceiling is not the train, it is the platform’s loading rate. A freight platform loads through belts, so the belt feeding it sets the throughput. Two Mk.6 belts into one platform gives 2400 items/min; two Mk.5 belts gives 1560. Any combination of marks works, and the sum is your actual rate.

This is the part worth internalising: upgrading your locomotive or adding cars does not increase throughput if the platform is fed by a Mk.3 belt. The bottleneck moves to whichever link is slowest, exactly as it does anywhere else in the factory.

When a train beats a belt

Trains win on distance and on volume-per-unit-of-effort. A single Mk.5 belt across 2 km costs a great deal of Alclad Aluminum Sheet and delivers 780/min. A train across the same distance costs track plus two stations and can deliver far more, provided you feed the platforms adequately.

Trains lose on latency and on complexity. Items arrive in batches, not continuously, so downstream machines need buffers to smooth the gaps. And a rail network has failure modes — signalling deadlocks, power spikes — that a belt simply does not have.

The honest split:

• Under about 500 metres: belt it. Simpler, continuous, no signalling.
• 500 m to a few km with high volume: train.
• Long distance, low volume: consider drones instead; they need no track.
• Fluids over distance: trains with fluid freight platforms, or pipes if the terrain is kind. Pipes lose to head lift on climbs; trains do not care about gradient for throughput purposes.

Building a line that will not need rebuilding

Run track on foundations, not terrain. Foundation-supported track is level, predictable, and easy to extend. Track laid on raw ground follows every bump and becomes the gradient problem described above.

Build both directions from the start or commit to a single-track loop with proper passing sections. Retrofitting a second track to a line built tight against terrain usually means rebuilding it.

Leave station platforms room to grow. Adding a third freight platform later is easy if you left space and awkward if you built a wall against it.

Signal as you build, not afterwards. Retro-signalling a finished network means reasoning about the whole thing at once, which is exactly when people put block signals in junctions.

Station layout: load and unload on the same platform

A freight platform does one job per train visit, set by the station’s schedule entry: load, or unload. The platform hardware is the same either way.

This means a two-station shuttle needs a platform at each end, and a longer route visiting three sites needs a decision at each stop about what happens there. Getting this wrong produces a train that circles endlessly moving nothing, because every stop is set to load and the cargo never leaves.

A layout that scales well: put the station at the end of a spur rather than on the main line. Trains pull in, do their work, and reverse out without blocking through traffic. Stations placed directly on a busy main line force every other train to wait.

Fluids on rails

Fluid Freight Platforms cost the same 50 MW and work the same way, with one extra consideration: fluid loading is rate-limited by the pipe feeding the platform, exactly as solid loading is limited by the belt.

The case for moving fluids by rail rather than pipe is elevation. Pipes lose throughput to head lift on every metre they climb, and recovering that needs pumps which cost power and space. A train does not care about gradient for throughput purposes — it costs more power to climb, but the amount delivered per trip is unchanged.

So the decision is roughly: flat ground and moderate distance favours pipes; significant climb or long distance favours rail.

Common mistakes