Factorio Forums

Frightning wrote:Why necro a 2-year old thread?

Technically, you can't "necro" a sticky thread .

siggboy wrote:

Frightning wrote:Why necro a 2-year old thread?

Technically, you can't "necro" a sticky thread .

How did I not noticed this was stickied?! D'oh!

Pigeon039 wrote:Trains are fun till you get ran over

Seconded. I was crossing tracks like two wagon's lengths ahead of a train (saw it coming, it drove slowly, I could easily make it)... then got killed by lag. Lost lots of resources I was carrying to craft the last gen armor. What a pain in the ass that was. Cost me an hour to gather enough goo, other shit, and craft it all again.

---

Anyway I was checking this thread out, hoping for a conclusion. Why do trains exist? If you have a very busy network with trains (without cargo wagons) stopping at every stop every 10 seconds, it's convenient for personal transport, but other than that (and that's not worth it) it has no advantage. Belts are not more expensive, less complex (even balancing), you need them anyway so you've got factories for them, and they don't require stone (which you'd otherwise not need in those quantities) or fuel.

The major advantage for trains is there speed.

Even carrying just one wagon, the train will be able to go long distances far faster then a belt could at far cheaper of a cost in resources. Trains are a more feast/famine system for your base, as when one shows up your network is SWAMPED with more ore then it can process. However when a train isn't there, all your feeding on is what little ore is left inside you base (if any).

I like to either do many 1 or 2 cargo wagon trains for ore, OR, one VERY large train (like 6 or 7 cars) of ore. With signals its very easy to build a system for just 2 or 3 trains so they don't ram into one another, even on a single rail line. That way you pretty much will always have at least one train waiting to unload, while one is unloading, and the 3rd is either getting loaded, or is also waiting (this is a good thing to have all 3 waiting).

With a single train running the ore, it is best to have a double Steel Chest system at the loading point, this makes sure that the chests are full of ore or at least are 1/2 full (when using 6x2 chests) making loading up shorter as then the train wont have to wait for fresh ore to hit the belt and get loaded. It does make the point a little bigger, but it gives a far bigger buffer then just using 6 chests and ensures the train will not be sitting long at the station.


Gotta feed that hungry hungry hippo called your base somehow!

Trains do two things better than anything else in the game: Travel fast, and move large volumes of items (high throughput). This makes them by far, the best system for moving large amounts of material between distant locations (I usually start considering trains if there is more than a few hundred tiles between where the stuff is, and where I want it to get to (with nothing in between that wants the stuff too). A single Cargo wagon can be loaded/unloaded by up to 12 inserters, which can be stack inserters for some truly extreme throughput. This can unload a full 2000 ore wagon in less than 6.3 seconds with full stack size upgrades! A train can have lots of wagons if you care to (though this comes at a cost in speed, which can be offset with additional Diesel locomotives). I love trains for feeding huge, resource hungry bases (that's tbqh, what their meant for, the tech is accessible fairly early to give a nice solution for getting scarce resources to your base from far away, e.g. very low resource settings makes trains really handy even in the early-mid game to get resources to your base cheaply and quickly).

If you generate a map with all resource settings maxed, then you are effectively sitting right on top of the resources, and trains will not be necessary for a simple playthrough.

If you build a large base that will mass produce rockets, and you don't game your settings that way, trains become mandatory at a point, because otherwise it's not possible to take the ore to the smelting in the amounts needed in a reasonable manner (drawing multiple blue belts over thousands of tiles is not "reasonable").

The best way to "force" trains in your games is to install the RSO mod. That will give you out-of-the-box settings for resource generation that mandate train use.

In a very simple playthrough (make 1 rocket, then quit), it's probably never necessary to build a train line at all (unless you force a very outlandish resource configuration).

I think belts should probably take power. Would fix this issue immediatly and makes sense 100%.

nuhll wrote:I think belts should probably take power. Would fix this issue immediatly and makes sense 100%.

i dont think its good, especially for the performance, when the game have to calculate power for 100k belts.

Why, power needs only to be updated when remove or add belts. They dont fall asleep.... like machines...

There is also a balancing issue:

If a belt consume 1 MW per tile, then you will simply add more Solar panels.

If you have a close range factory, where you need tons of belts, bots will become better at a point making belts obsolete.

Belts consuming energy will only slow down the early stage of the game, also you have the first belts without having energy.

It will be very hard to adjust the power of belts to keep them balanced.
Powered Belts could be the an idea for T4+ belts, that are faster then the yellow, red, blue belts. But instead of Belts, transport tubes.

I think energyx belts could be a think for a mod, but not for the vanilla game.

energy bots > belts > trains

nuhll wrote:energy bots > belts > trains

Logibots>belts is true imho, but trains are in a league of their own for long distance transport of large quantities of stuff, nothing else comes close, not even belts, not even bots.

Trains are FAR more flexible. They always pick the most efficient route (if you've signalled properly) and once you've laid the tracks you'll probably never need to improve it, just expand.

However I think this is mostly a theoretical discussion. In reality, the supply of the resource is generally the limiting factor. It's extremely simple to have 4 wagons, either in a single train or split between two, that pick up everything from an outpost faster than it can be mined. When throughput is capped by the supply rather than the logistics, the best means of transport then becomes what is cheapest and most convenient. The first is probably debatable but trains are more convenient hands down.

The only case throughput becomes a realistic factor is when you add speed modules or whatever to generate an absurd amount of ore, but then throughput is easily affected by these factors:

• Distance: To improve throughput of belt, you need to add more belts the entire length. You only need add an extra train for which length is irrelevant to the cost. Therefore when the cost of belts of that distance exceeds that of the train, the train is more cost efficient.
• Other resources: Again you need only one rail for everything, but when using belts you need dedicated belts for everything. This substantially increases the cost of belts per new resource.
• Sustainability: When an outpost is depleted, you just redirect the train to a new one and it can be reused. Belts need to be picked up for the length they will no longer be used, then laid down again, which consumes a lot of player time.

It's the original "belts work for shorter distances" argument. I mean you can obviously see that a train carriage moves faster than items on a belt, and it holds more per tile, so whatever way you look at it trains are faster. Sure they might be a bit more expensive but the difference is not great for the improved response time and throughput, and once you start expanding it becomes cheaper anyway.

Deadly-Bagel wrote:Trains are FAR more flexible. They always pick the most efficient route (if you've signalled properly) and once you've laid the tracks you'll probably never need to improve it, just expand.

However I think this is mostly a theoretical discussion. In reality, the supply of the resource is generally the limiting factor. It's extremely simple to have 4 wagons, either in a single train or split between two, that pick up everything from an outpost faster than it can be mined. When throughput is capped by the supply rather than the logistics, the best means of transport then becomes what is cheapest and most convenient. The first is probably debatable but trains are more convenient hands down.

The only case throughput becomes a realistic factor is when you add speed modules or whatever to generate an absurd amount of ore, but then throughput is easily affected by these factors:

• Distance: To improve throughput of belt, you need to add more belts the entire length. You only need add an extra train for which length is irrelevant to the cost. Therefore when the cost of belts of that distance exceeds that of the train, the train is more cost efficient.
• Other resources: Again you need only one rail for everything, but when using belts you need dedicated belts for everything. This substantially increases the cost of belts per new resource.
• Sustainability: When an outpost is depleted, you just redirect the train to a new one and it can be reused. Belts need to be picked up for the length they will no longer be used, then laid down again, which consumes a lot of player time.

It's the original "belts work for shorter distances" argument. I mean you can obviously see that a train carriage moves faster than items on a belt, and it holds more per tile, so whatever way you look at it trains are faster. Sure they might be a bit more expensive but the difference is not great for the improved response time and throughput, and once you start expanding it becomes cheaper anyway.

With the right map settings, even then, even yellow belts are still more expensive. If you make the map have Very low frequency but very big and very rich deposits, then you can get an average of roughly 80 miners (I've nearly had 130 on a modestly larger than average copper deposit), and this is a 'coverage' layout rather than max density (which can only be achieved by bots). Without modules, ore miners produce 0.525 ore/sec (except stone, at 0.65/sec), so a single blue belt can handle about 75 miners with at full compression, which means 3 yellows is tentatively enough for a single resource deposit. Now, said belts cost 1.5 Iron/tile, so each tile of distance you need to traverse is roughly 4.5 iron a tile.

Train rail costs 1 steel plate, 1 iron stick, and 1 stone, for 2 rails that cover 2 tiles each, for a cost of 1.5 iron (counting steel is iron), and 0.25 stone per tile, which is considerably less. Now, fixed costs are also higher with trains, especially if you build for max throughput (e.g. 12 steel chests=96 steel=480 iron, need those 2x per wagon, though unload might service several mines, lowering impact to fixed cost/resource patch, but also, stack inserters; up to 24/wagon per end of route if you load/unload from belts, 12 per end with bots).

For these reasons, shorter distances are often much better to use belts (think less than say, about 200 tiles). But after you hit about 200 tiles, it starts to make more sense to seriously consider setting up trains (the distance based variable cost starts to become pretty significant by then).

Hadn't bothered to do the calculations until now, 5.5 iron and 1 stone per 2 tiles for rail, and 23 iron per 2 tiles for red belts. A train and two carriages is 15 copper and 560 iron, so you only have to go 64 tiles to break even in terms of iron. This doesn't include signals but that won't affect it much. Yellow belts might technically be cheaper but their lack of throughput makes them a poor choice.

Deadly-Bagel wrote:Hadn't bothered to do the calculations until now, 5.5 iron and 1 stone per 2 tiles for rail, and 23 iron per 2 tiles for red belts. A train and two carriages is 15 copper and 560 iron, so you only have to go 64 tiles to break even in terms of iron. This doesn't include signals but that won't affect it much. Yellow belts might technically be cheaper but their lack of throughput makes them a poor choice.

Rail recipe makes 2 Rail, which are each 2 tiles long, meaning the recipe covers a distance of 4 tiles. Also, for fixed costs, you need to account for much more than just wagons (which is a throughput variable cost). You also need to account for inserters, and perhaps buffer chests. So the break even point is a lot higher than 60. Also, yellow belts are FAR cheaper per unit distance than red belts (1.5 iron/tile instead of 11.5 iron/tile).

Well yes but a single two-way rail isn't particularly expandable so that accounts for rails going both directions. There is also the cost of signals, two train stops and having the rail loop around so it's not an exact figure but it gives you an idea. From this we know the break-even point isn't going to be 1,000 tiles or something.

Also yellow belts are 3 iron per two tiles, which is a bit over half the cost of the rails. However the poor throughput makes it a viable option only for coal until plastic starts seeing heavy use, or perhaps stone, but typically I find I don't get through my starting deposits or either until much later in the game, when I already have a rail network. So then we can calculate cost like this:

A train (including wagons) is worth approximately 192 yellow belts (assuming 1 copper = 1 iron). For simplicity let's assume 1 rail pair (two-way) is worth four yellow belts (stone = 0.5 iron). Therefore the break-even point becomes roughly where the outpost is 292 tiles away from the base plus double the distance between the outpost and the nearest rail.

For example if the nearest stone deposit is 600 tiles away, but has a rail running 100 tiles from it, it's cheaper to make the train and rails than it is to make the belts because you only need to link rails to the network but belts need to be run the full length. Granted, this is a fair distance to start breaking even but the ludicrous difference in throughput is well worth it.

Deadly-Bagel wrote:Well yes but a single two-way rail isn't particularly expandable so that accounts for rails going both directions. There is also the cost of signals, two train stops and having the rail loop around so it's not an exact figure but it gives you an idea. From this we know the break-even point isn't going to be 1,000 tiles or something.

Also yellow belts are 3 iron per two tiles, which is a bit over half the cost of the rails. However the poor throughput makes it a viable option only for coal until plastic starts seeing heavy use, or perhaps stone, but typically I find I don't get through my starting deposits or either until much later in the game, when I already have a rail network. So then we can calculate cost like this:

A train (including wagons) is worth approximately 192 yellow belts (assuming 1 copper = 1 iron). For simplicity let's assume 1 rail pair (two-way) is worth four yellow belts (stone = 0.5 iron). Therefore the break-even point becomes roughly where the outpost is 292 tiles away from the base plus double the distance between the outpost and the nearest rail.

For example if the nearest stone deposit is 600 tiles away, but has a rail running 100 tiles from it, it's cheaper to make the train and rails than it is to make the belts because you only need to link rails to the network but belts need to be run the full length. Granted, this is a fair distance to start breaking even but the ludicrous difference in throughput is well worth it.

I've yet to encounter a situation where I am over-taxing my mixed rail system (mostly 2-way rail, except at the outposts and unloading stations where I used 1-way for simplicity...trains are one-way trains), that includes my 8000 plates/min kilobase, the rails connecting my 3-bay unloading system to my mining outposts are 2-way rails. So unless you're building very big (megabase territory), or going full-towns mode (1 process per tower with few exceptions) you probably won't have issues.

I have an idea for train / vonveyors issue.

For each 20 belts long 20 % less belts speed.
21 belts 80 % speed
41 belts 60 % speed
61 belts 40% speed
81 belts 20% speed
100+ belts 10% speed

it would be totally unacceptable for very long resources gathering and then trains win.

Additionally I will lower starting speed and acceleration parameters of the train and speed those up by research.

Or perhaps for each 40 belts ? Or simply if belts are 100+ long speed = 10 %

Predator wrote:I have an idea for train / vonveyors issue.

For each 20 belts long 20 % less belts speed.
21 belts 80 % speed
41 belts 60 % speed
61 belts 40% speed
81 belts 20% speed
100+ belts 10% speed

it would be totally unacceptable for very long resources gathering and then trains win.

Additionally I will lower starting speed and acceleration parameters of the train and speed those up by research.

Or perhaps for each 40 belts ? Or simply if belts are 100+ long speed = 10 %

There's no reason for a belt to slower just because it's longer, and what about T-junctions and splitters, now how long is your belt?