Factorio Forums

Hi there!
So I've got a quite large base which is divided by two big parts, running simultaneously: main base with research labs and science packs production, and spaceships (autolaunch) part to supply main base with space science packs.
Main activities related to expansion are finished but I still need to expand, even if it's not so drammatically.

And and I have a bottleneck in iron/copper plates production. The part which I need advice on.

Currently I have following logistics:
- All trains with Iron arrive at ONE receiver station, copper trains arrive to another. - As you can see, every station has logistics network and robots quickly unload trains to the buffer of storage chests.
- As you can see also, there are transporter belts with requester chests, that are filled by the same robots with copper/iron so it goes by belts to furnaces.
- Every belts line is compressed maximally and processed by furnaces in the way that the output is fully compressed belt of iron/copper plates. - Also, every station actually is doubled because I have two train stops with same name. This allow trains to wait minimum in a waiting pool and unload in almost two times faster, increasing input of raw materials.

The issue is that there so many belts of raw material going from stations to furnaces, that their consumption is bigger than input even from two train stops constantly occupied by fully loaded trains.

So input of raw iron/copper from mines to main base is going too slow to cover all my needs and load all furnaces to 100%.

That's the issue that I want to solve.
What I tried:
- Put production modules there possible and reasonable. This is done. So I'm lowering raw material consumption where I can. Electricity is not a problem, I like nuclear stations and have 5 sections x 8 reactors each with 5 megawats in total.
- Remove unnecessary belts combining some optional staff which produced rear (turrets, roboports, robots, etc.) with highly loaded belts. This is compromise but simplifies raw material logistics.
- Add more train stops. Doesn't work as too much distance for robots to unload it in an efficient way, see screenshot 1.
- Add more storage. Doesn't work for the same reason. Plus storage amount increasing is short-term solution which will not help in a long run.
- Use one part of the base at a time - main OR spaceships. Is not good because endless research goes dramatically slow overall in this case.
- Rebuild furnace lines inserting two production modules in every one. I expect increasing plates production by 30% by doing it but practically it is less due to these bonus plates are hard to maintain between belts and belts get over-compressed or under-loaded. Maybe you can suggest some solution here.
- Making more receiving train stations with different names. This may be a solution however I have such amount of trains that It looks just completely insane to maintain (redirect) trains between different receiving stations (for example, when some mines depleted or new are built).

So, appreciate any advises on how can I avoid this bottleneck,
maybe some examples of economy production circuits, or furnaces upgrades, or receiving stations refactoring.

I'm playing no mods last version of the game which I like, so modding is not an option, taking game performance in account as one of the reasons.

Thanks at lease that you could read this post entirely! And have a nice day!

What you need to do is abandon belt felt designs for your smelty and go with speed beacon'd prod moduled smeltery. Have your trainstop as close to the furnace block as possible (don't bother buffering with storage chests, just let the requestor chests at each furnace serve as your buffer (2400x# of furnaces=plenty sufficient ore buffer). A proper beaconized layout will have 8 beacons affecting each furnace and as it gets larger, will get closer to a 1-to-1 beacon to furnace ratio (starts 8-to-1 and decreases as you extend rows down towards 2-to-1 ratio, adding additional rows pushes it down closer to 1-to-1; realistically, there is a limit due to roboport logistics area limiting things in at least one direction). This layout will result in (with all modules at tier 3) +370% speed with +20% productivity at the smeltery stage, so it's extremely compact while providing massive throughput, and the logistics bots and trains can totally keep up with this kind of setup.

Edit: Upon reflecting a bit more on your situation, I realized the above doesn't address your current problem (though it's still a good thing to do for the efficiency of your setup). Essentially your problem is lack of train throughput, and you are currently running 2-6-0 if I'm seeing that right. Long story short, the solution is to run longer trains (with more locos if needed for performance). The way to do this to where it can scale basically indefinitely is to create a 'hub' area where all your copper ore is collected to. This will (for now) be fed straight from your copper mines by your current 2-6-0 trains. The hub has outgoing trains to your copper smeltery which are longer (maybe 4-20-0?) which are, in effect, fed by your more numerous, smaller trains. The hub is just a set of train stops with a ling of roboports with logistics bots between them, with the bots being responsible for moving the ore from the small trains to the long ones. This concept can be iterated on too, where you have small trains feeding medium sized trains (at multiple smaller hubs) that all fed into a larger hub, etc. (as many layers as needed in principle).

Thanks for the response!
Both part of your post are extremely helpful! So thanks a lot for such clear explanation!

As for the 1st part - my furnaces are very close to unload stations however because there are a lot of them, some are quite distant. However with implementing of your design I can see this will not be an issue. So this is definitely what I will do now.

Second part is more interesting. I already had an Idea to make a hub, the main thing with it is that hub increases time raw materials are delivered to the base, I mean their "delivery latency".
However, on the current level, it's not a major issue anymore so thank you very much for bringing this idea!
UPD: also thinking now of adding smelting to the hub, however it contradicts with the main hub goal ))) so need to think more.

Also I have issues in the base with dimensions of receiver stations. I'm somewhat limited by height and width of the area which can receive trains, so I'm not sure that long trains will fit even after furnace upgrade.
However it's possible so definitely I'll try to improve my current system as I can using your ideas.

Thanks again!

There's a hard limit to how fast you can unload a single wagon, so at certain stage, "add more trains/wagons" becomes the only solution for this kind of problem. However there are things you can do to get closer to that limit.

1. Research stack inserter bonus and train breaking distance. First improves unloading speed directly, second greatly decreases time that the station stays unoccupied.
2. Setup your waiting bay so that another train waits for directly behind the currently unloading one (this seems to be the case for the top train on your pic).
3. Place intermediate regular signals between wagons on the unloading station. This ensures that the next train starts moving into the station after departing train has moved mere 7-8 tiles, instead of full length. You've got lights on your station. Those lights should be signals.
4. Optimize your trains loco-to-wagon ratio. Make it 1:2 and don't use bi-directional trains. This is again to improve train acceleration/deceleration performance and thus decrease station downtime between trains.

And finally, an unexpected advice:

5. Use shorter trains.

Consider this: when you have a 10-20-0 unloading station, station downtime between trains is equivalent to the time that the next train needs to cover that much distance - i.e. 210 tiles or so. It can be reduced with train breaking distance research, but it is still a considerable time. When you have a 1-2-0 unloading station however, station downtime is equivalent to the time that the train needs to cover mere 21 tiles or so. And when you have ten stations of 1-2-0 design, the entire downtime of your stations system is the same "move 21 tiles" equivalent. Downtime won't be 10 times smaller in this example, as train needs to accelerate and decelerate from zero, but it will still be considerably less.

Of course, there's another limit that you'll hit and that's the train throughput ability of your rail network. So using 1-2-0 trains for your most throughput critical deliveries might be slightly unwise. 2-4-0 or 3-6-0 though might be a good compromise. Do not go for behemoth trains unless you really have no choice.

Lav wrote: 5. Use shorter trains.

Consider this: when you have a 10-20-0 unloading station, station downtime between trains is equivalent to the time that the next train needs to cover that much distance - i.e. 210 tiles or so. It can be reduced with train breaking distance research, but it is still a considerable time. When you have a 1-2-0 unloading station however, station downtime is equivalent to the time that the train needs to cover mere 21 tiles or so. And when you have ten stations of 1-2-0 design, the entire downtime of your stations system is the same "move 21 tiles" equivalent. Downtime won't be 10 times smaller in this example, as train needs to accelerate and decelerate from zero, but it will still be considerably less.

Of course, there's another limit that you'll hit and that's the train throughput ability of your rail network. So using 1-2-0 trains for your most throughput critical deliveries might be slightly unwise. 2-4-0 or 3-6-0 though might be a good compromise. Do not go for behemoth trains unless you really have no choice.

If you have a 10-20 and a 1-2 station, trains might get from the wwaiting area to the 1-2 station 10x as fast, but it will need to happen 10x as frequesntly as well - I'd say there isn't much difference.

Jap2.0 wrote:

Lav wrote: 5. Use shorter trains.

Consider this: when you have a 10-20-0 unloading station, station downtime between trains is equivalent to the time that the next train needs to cover that much distance - i.e. 210 tiles or so. It can be reduced with train breaking distance research, but it is still a considerable time. When you have a 1-2-0 unloading station however, station downtime is equivalent to the time that the train needs to cover mere 21 tiles or so. And when you have ten stations of 1-2-0 design, the entire downtime of your stations system is the same "move 21 tiles" equivalent. Downtime won't be 10 times smaller in this example, as train needs to accelerate and decelerate from zero, but it will still be considerably less.

Of course, there's another limit that you'll hit and that's the train throughput ability of your rail network. So using 1-2-0 trains for your most throughput critical deliveries might be slightly unwise. 2-4-0 or 3-6-0 though might be a good compromise. Do not go for behemoth trains unless you really have no choice.

If you have a 10-20 and a 1-2 station, trains might get from the wwaiting area to the 1-2 station 10x as fast, but it will need to happen 10x as frequesntly as well - I'd say there isn't much difference.

Except time to leave station is not linear in length of train because a train with the optimal ratio of locos to wagons accelerates at the same rate regardless of how long it is. This means that it's average speed over the act of leaving the station is higher than a shorter train with the same performance. Someone else made the brilliant suggestion to put signals between the wagons (and locos) at the station so that the next train can begin pulling as soon as the previous train has cleared one wagon worth of distance (7 tiles), this lets you really take advantage of what I have pointed out just now.

Frightning wrote:

Jap2.0 wrote:

Lav wrote: 5. Use shorter trains.

Consider this: when you have a 10-20-0 unloading station, station downtime between trains is equivalent to the time that the next train needs to cover that much distance - i.e. 210 tiles or so. It can be reduced with train breaking distance research, but it is still a considerable time. When you have a 1-2-0 unloading station however, station downtime is equivalent to the time that the train needs to cover mere 21 tiles or so. And when you have ten stations of 1-2-0 design, the entire downtime of your stations system is the same "move 21 tiles" equivalent. Downtime won't be 10 times smaller in this example, as train needs to accelerate and decelerate from zero, but it will still be considerably less.

Of course, there's another limit that you'll hit and that's the train throughput ability of your rail network. So using 1-2-0 trains for your most throughput critical deliveries might be slightly unwise. 2-4-0 or 3-6-0 though might be a good compromise. Do not go for behemoth trains unless you really have no choice.

If you have a 10-20 and a 1-2 station, trains might get from the wwaiting area to the 1-2 station 10x as fast, but it will need to happen 10x as frequesntly as well - I'd say there isn't much difference.

Except time to leave station is not linear in length of train because a train with the optimal ratio of locos to wagons accelerates at the same rate regardless of how long it is. This means that it's average speed over the act of leaving the station is higher than a shorter train with the same performance. Someone else made the brilliant suggestion to put signals between the wagons (and locos) at the station so that the next train can begin pulling as soon as the previous train has cleared one wagon worth of distance (7 tiles), this lets you really take advantage of what I have pointed out just now.

So then if you spam signals in the station, due to the increased amount of acceleration would longer trains be faster?

Jap2.0 wrote:

Frightning wrote:

Jap2.0 wrote:

Lav wrote: 5. Use shorter trains.

Consider this: when you have a 10-20-0 unloading station, station downtime between trains is equivalent to the time that the next train needs to cover that much distance - i.e. 210 tiles or so. It can be reduced with train breaking distance research, but it is still a considerable time. When you have a 1-2-0 unloading station however, station downtime is equivalent to the time that the train needs to cover mere 21 tiles or so. And when you have ten stations of 1-2-0 design, the entire downtime of your stations system is the same "move 21 tiles" equivalent. Downtime won't be 10 times smaller in this example, as train needs to accelerate and decelerate from zero, but it will still be considerably less.

Of course, there's another limit that you'll hit and that's the train throughput ability of your rail network. So using 1-2-0 trains for your most throughput critical deliveries might be slightly unwise. 2-4-0 or 3-6-0 though might be a good compromise. Do not go for behemoth trains unless you really have no choice.

If you have a 10-20 and a 1-2 station, trains might get from the wwaiting area to the 1-2 station 10x as fast, but it will need to happen 10x as frequesntly as well - I'd say there isn't much difference.

Except time to leave station is not linear in length of train because a train with the optimal ratio of locos to wagons accelerates at the same rate regardless of how long it is. This means that it's average speed over the act of leaving the station is higher than a shorter train with the same performance. Someone else made the brilliant suggestion to put signals between the wagons (and locos) at the station so that the next train can begin pulling as soon as the previous train has cleared one wagon worth of distance (7 tiles), this lets you really take advantage of what I have pointed out just now.

So then if you spam signals in the station, due to the increased amount of acceleration would longer trains be faster?

A single station with longer trains will have better throughput than with shorter trains, but it is also true that 10x 1-2-0 stations will outperform a single 10-20-0 station (*provided you can handle the traffic sufficiently well that it isn't causing a bottleneck of its own*). This is because the downtime is less for shorter trains and you have the same number of wagon unload spots in both setups. In other words, it can be better parallelize with multiple stations and trains a bit before going to lengthening them as a means of increasing throughput, but eventually you will need to exploit both (because train networks can only handle so much traffic in a problem-free manner).