Understanding the Main Bus
The main bus is just a wide conveyor line with splitters feeding recipes along the side. You pour your raw resources into one end, and each recipe pulls what it needs from its own lane. Everything else is layout and lane management. Most beginners spend weeks building inefficient factories before they even learn this pattern. Then they find out it takes about twenty minutes to set up once you know the dimensions. A standard four-split main bus has four output lanes per input. Each lane feeds a single recipe. You align your assemblers perpendicular to the bus so input belts connect from below and output belts connect from above. That's the basic shape. From there it's just stacking more and more lanes as your production needs grow. There are really two main variants: the two-split and the four-split. A two-split bus uses a 2x2 splitter and gives you two output lanes per input. This is what most people start with because it's compact and easy to route. A four-split bus uses a 2x4 splitter and gives four output lanes. It uses more splitters and takes up more horizontal space, but it delivers more items per second through each output because the splitting loss is lower.
The two-split Loses about 25 percent of input throughput per splitter stage when feeding from one side. A four-split loses about 12.5 percent under the same conditions. That difference matters when you're pushing near belt limits. If you're making ten iron plates per second and ten iron gears per second on a two-split, you need four input lanes of iron ore just to cover the splitter loss. On a four-split you only need three. This is the kind of detail nobody mentions until you're staring at a starved assembler and wondering where your belts went wrong.
Building the Bus: Step by Step
Start with your most basic resource. Iron is the usual choice since everything eventually depends on it. Lay down four lanes of fast transport belt as your input. Each lane carries 60 items per second, so four lanes give you 240 items per second entering the bus. That sounds like a lot until you start splitting it out. Place a 2x4 splitter at the start of your bus section. Connect your four input lanes to the left side. The right side now has four output lanes. Each output lane runs parallel to the bus and feeds one recipe. Put an iron plate assembler on the first lane, a gear assembler on the second, a copper plate assembler on the third, and a copper gear assembler on the fourth. Route the input belts from below the bus and the output belts from above. That's one complete section. When you need more recipes, add another splitter section downstream. Don't extend the same bus further than about eight to ten recipes before adding a new splitter block. After that many recipes the belt routing gets messy and you start fighting with your own construction. I've seen players try to run a single bus past twenty recipes and then spend hours reworking their entire layout because the output belts collide with each other.
Get the Full Details
Advanced Routing: Mixing Resources
One thing that catches people off guard is that you can run different resources on the same bus. The bus doesn't care what the items are. It only cares about belt throughput. So you can have one lane carrying iron ore, the next carrying copper ore, the next carrying iron plates, and the next carrying steel. The splitters will divide whatever is on each lane independently. This is how you build a combined bus that feeds both iron and copper products without running separate infrastructure for each resource. A counter-intuitive point about this is that mixing resources on a shared bus actually improves throughput efficiency in some cases. When you run separate buses for iron and copper, each bus has its own splitter losses. When you combine them on one wide bus, you share the splitters. You still lose the same percentage per splitter, but you use fewer splitters overall. The trade-off is that your bus gets wider and harder to route around existing factories.
A Real Problem I Ran Into
Last week I was building a bus for a mid-game setup with iron plates, iron gears, copper plates, and steel. The iron gear assembler kept stalling even though I had plenty of iron ore going in. I checked the belts three times. Everything looked correct on paper. The input had six lanes of iron ore feeding into a 2x4 splitter, which should have given me three lanes of iron ingots to each recipe. It didn't. The issue was that the iron plate assembler was pulling two full lanes of output from the same splitter, leaving only two lanes for the gear assembler. But each gear needs six plates and six ingots, so the plate consumption was starving the gear input. I solved it by adding a second 2x4 splitter downstream specifically for the gear lane, pulling directly from the raw ore bus instead of from the processed ingot bus. The gear assembler stopped stalling immediately. The fix cost me two splitters and about a meter of extra bus length.
Common Pitfalls
The biggest mistake is assuming that belt count equals recipe capacity. A ten-lane bus doesn't mean you can run ten recipes efficiently. Each recipe needs its own input lane and its own output lane. If you're making complex products that require multiple inputs, you need multiple lanes per recipe. A single circuit condition or train stop can choke an entire lane. I've lost hours debugging a bus only to find one broken splitter caused a cascading starvation across six recipes. Another pitfall is not accounting for future expansion. Build your bus with extra lanes from the start. A bus that's exactly wide enough for your current recipes will need to be completely rebuilt when you add your twentieth recipe. Adding two empty lanes costs nothing during construction but saves you from tearing down and rebuilding later. The same logic applies to your input infrastructure. Five input lanes are easier to add early than mid-game when your factory is already built out.
When the Main Bus Fails
The main bus has hard limits. It doesn't scale well beyond roughly fifteen to twenty recipes on a single bus section. After that you're spending more time routing belts than producing anything. It also doesn't handle products with wildly different throughput requirements well. If one recipe needs five items per second and another needs two hundred, they'll fight for space on the same bus and one will always be starved. For large-scale production, you're better off splitting into sub-buses or using a combined bus with dedicated lanes per product family. A steel production line that needs hundreds of iron plates per second should have its own dedicated bus section, not share lanes with gear assemblers making a handful of products per minute. The main bus is a great pattern for the early and mid game. It falls apart under heavy industrial demand. That's not a flaw in the design. It's just what happens when you try to force a simple pattern into a complex problem.
Practical Numbers to Keep in Mind
Fast transport belt moves 60 items per second. A 2x4 splitter fed from one side gives you 75 percent of input throughput on each output lane. A 2x4 splitter fed from both sides gives you about 93 percent. If you're pushing more than 120 items per second through a single output lane, you need two lanes minimum, and you should probably be questioning whether that recipe belongs on the main bus at all. Underground belts add a small throughput penalty of roughly 2 to 3 percent per underground due to item pathing. Don't worry about it unless you're optimizing for maximum throughput. It won't make or break your factory.
The Bottom Line
Build a four-split main bus with four input lanes and eight output lanes. Add one recipe per output lane. Leave two empty lanes on each end for expansion. Route inputs from below and outputs from above. Don't exceed ten recipes per bus section without adding a new splitter block. Use sub-buses for high-throughput products. That's the whole pattern. Everything else is just adjusting the numbers to fit your factory size.
