Belts, how far is too far?

[MeduSalem]

I'm a little worried that the belt optimizations in 0.16 will make this the most UPS-efficient delivery method too. I don't want to see sixty blue belts replacing one train line.

Just because something is UPS-efficient doesn't mean it's the only way to play the game.

Okay, maybe if you build a huge mega base with several hundreds of hours invested that only runs at 4-5 FPS anymore... but I doubt that this is the case for most players.


At least I wouldn't feel like stamping a million blue belts into the landscape. At some point a rail network becomes just easier to use and expand... and also nicer to look at in my opinion.

But that said I don't think that belts will be the most efficient method even with the 0.16 optimizations.

[Koub]

People usually start to worry about UPS when they feel the UPS wall as their base gets too big for their computer to handle.
However, there are a LOT more train users than there are people suffering from UPS issues.

There is no such thing as a bad optimization (as long as no bugs are involved).

[quyxkh]

Well, my preferences don't match Elok's either, but I think he's got his facts straight. A 10-wagon trainstop is 30 belts if I've got it right (I think the theoretical limit is 10 wagons × 12 inserters/wagon × 12items/swing ×60ticks/sec / (26ticks/swing + 12 items×9 slots/item / 3 slots/sec)/40 items/sec/belt==34.8387 belts if I typed the units in right), I'm very sure I'm not the best at this but my attempts so far at lane-balancing not-small trains have been uuuuuu-glY.

There has to be a better way than this to balance unloading 10 wagons to 30 belts, right?

screenshot.jpg (1.26 MiB) Viewed 5443 times

I was wondering how belt-fed factories really stack up against bot-fed factories. In isolation, where you don't have to feed and clean up after them, they're very competitive, but I don't know how to make them play well with trains. I don't have a good test rig for no-trains benchmarking yet and I like trains anyway so this is as far as I've got.

I kinda thought Loewchen said it best up front, it's all personal preference. Did that need repeating? Anyway, when you hit KSPM-ish territory, yeah, UPS is going to start being an issue, not much of one for a while even after that tho. All the same, UPS issues or no, the idea of using sixty belts to replace a train line repels me.

[Frightning]

Beyond a few hundred tiles, trains are always cheaper than belts for a non-trivial amount of throughput, and they scale up much better than belts too.

[PunkSkeleton]

No matter how much you twist it, train add turbulence in the supply chain.

Not only your unloading must be way faster than your intake because of the downtime between train, but there's the logistic of train that don't always scale well (adding more train raise the odds of stopage and traffic) unless you have a dedictated rail for each of your mine.

When you use conveyor, a full conveyor 100 miles away will stay full until it reach it's destination. So it's only a matter of making sure you mine enough ore to fill them up.

When you load a train it will also stay full until it reaches it's destination... Belts don't scale well because of the space required.

Taking my base in example, I got a conveyor highway with my 9 line of blue conveyor (5 for iron, 4 for copper) at 100% capacity. And it scale well so I could raise it to 20, 50, 100 easilly (It's a matter of a few copy&paste).

And running back and forth for more blue belts because your inventory is limited. If I often run out of belts when building mining outpost then I can't imagine what would happen if I would have to put 1.5 km of blue belt for each blue belt of ore (and my stations usually produce more than one blue belt).

The only, only real downside of conveyor compared to train is the increased cost of iron.

No, the real downside of belts are:
1. Space required.
2. Lack of flexibility when expanding - when your mining outpost starts to run out of ore you have to add a new outpost nearby or at least in the same direction from the main base or your whole giant stretch of blue belts will be wasted. When my mining outpost starts to run out of ore I don't have to do anything (I can deconstruct it once it's completely dry but I don't have to). I only have to look at my main buffer, that's the only place of interest for me. If it's running low then it is nice to add a new station which will grab the ore from a few new mining outpost. I can do it in any place on the map.
3. Time required to expand the throughput of long lines.

Nice base. You should make a video out of it. If you got Windows 10, there's a great video recorder included : https://gizmodo.com/windows-10-is-hidin ... 1719196149

Nothing special about it IMO and people would get bored travelling around a huge base at 40 UPS...

[featherwinglove]

However, there are a LOT more train users than there are people suffering from UPS issues.

Yeah. There's a reason for that. Probly same reason there's not a lot of overlap in those groups

[malventano]

Any numbers on pipes loosing pressure?

Sure: 19851

The figures from that post don't apply anymore. Currently, flow levels out at ~1,000 units/second through 200 pipe tiles (or 100 underground pipe tile pairs), for a maximum distance of 1,000 tiles, before a pump is needed.

Due to the flow/level/pressure falling off more sharply after the first few pipe tiles, getting higher flow rates over longer distances runs into an issue of it being impossible to 'fan out' from the flow source while still maintaining the same distance possible with a single pipe run. That's what prompted me to create the Manifolds mod, which enables full 12,000/sec pump flow across parallel pipe sections, spanning 1,000 tiles before another pump is required.

[Elok]

The only, only real downside of conveyor compared to train is the increased cost of iron.

No, the real downside of belts are:
1. Space required.
2. Lack of flexibility when expanding - when your mining outpost starts to run out of ore you have to add a new outpost nearby or at least in the same direction from the main base or your whole giant stretch of blue belts will be wasted. When my mining outpost starts to run out of ore I don't have to do anything (I can deconstruct it once it's completely dry but I don't have to). I only have to look at my main buffer, that's the only place of interest for me. If it's running low then it is nice to add a new station which will grab the ore from a few new mining outpost. I can do it in any place on the map.
3. Time required to expand the throughput of long lines.

1.
Taking this example : download/file.php?id=10829

This allow 4 train track which if commonly refered as the "sweet spot" since adding a train raise the number of train exponentialy. I've yet to see a base that can fully use 4 train tracks.

The whole thing is 16 space wide. Meaning that it could be replaced by 16 belts track. Granded the train have a higher volume, but at 16 blue belt it'll take a loooooong time to fill them up completely. In fact, in a normal game with 'normal' ressource I doubt you'll have the time to fill them up since you'll bassically need 16 mine running at the same time.

2.
Everything you just said is plausible for both train and belt. I could abanddon a dried mining base too (or course, it's more worth it to send my drone to recuperate everything when I'm using belt). If you send a train of belt "highway" to a direction where there's no more ressource and you end up in a big lake, the whole thing is wasted. What's different between laying belt or laying train tracks?

3.
I think I just understood that you'll placing everything by hand. I use drone all the time to do everything (They also place their Drone base to expend). I have a Bluepring of my 16-blue belt highway with drone base, power and laser defence that measure 200 block and is repeatable. If I want to expand more to the East, I only need to paste my blueprint a few time and the drone will do the job. Expanding to a new mining base take about 1 min of my time.

Of course my drone will do the job a lot faster if they had to place 4 train tracks instead of 20 belt track. But they are drone, they don't complain.

[SuicideJunkie]

1)
Yep. Belts have lower throughput and far more lag end to end.
And you can't use the belt for other types of shipments.

2)
Rails are omnidirectional.
You can run trains back up the track, and put a factory where the ore patch was without changing anything except the stops. The track will also continue to be used as an alternate route through your base if necessary.

3)
No. Why would you even consider this valid for a second?
Using drones is just as easy, but faster and cheaper with rails.
Rails are so extremely cheap that you can even roll along at a decent clip in a train and use your personal roboport to lay long track without needing infrastructure.


Personally, I like the taxi service that trains provide. Blasting me around my base at 300km/h to places I only know the name of but not how to get there.
I haven't yet bothered with building past red belts, since there's no need for extra speed along a couple tiles worth of belt. Bots and trains mean yellow belts continue to be sufficient for the niche use of scootching parts a couple of tiles although the extra reach of red underground belts has come in handy.

[EpicFail1403]

The whole thing is 16 space wide. Meaning that it could be replaced by 16 belts track. Granded the train have a higher volume, but at 16 blue belt it'll take a loooooong time to fill them up completely. In fact, in a normal game with 'normal' ressource I doubt you'll have the time to fill them up since you'll bassically need 16 mine running at the same time.

16 blue belts = 16 * 40 items/s = 640 items/s
which is NOT much. Note that this 640 item/s includes all your iron ore and copper ore and coal and stone and uranium ore. Reaching this state doesn't prove that belts scale up well enough.

[Frightning]

The whole thing is 16 space wide. Meaning that it could be replaced by 16 belts track. Granded the train have a higher volume, but at 16 blue belt it'll take a loooooong time to fill them up completely. In fact, in a normal game with 'normal' ressource I doubt you'll have the time to fill them up since you'll bassically need 16 mine running at the same time.

16 blue belts = 16 * 40 items/s = 640 items/s
which is NOT much. Note that this 640 item/s includes all your iron ore and copper ore and coal and stone and uranium ore. Reaching this state doesn't prove that belts scale up well enough.

My 1-track wide, 1-way rail system in my kilobase back in 0.13 days could handle 8k iron and 4k copper ore per minute plus some stone and coal (power was from solar though) with just 6 1-2-0 trains (2 of which didn't run frequently because demand for Coal and Stone was low). Such a rail network is orders of magnitude cheaper and easier to setup, and to expand as needed.

[Zavian]

Well even a "small" post rocket base which only does 600 science per minute needs about 30k iron ore + 30k copper ore per minute these days. So the 12k total from that 8k iron + 4k copper network would be enough throughput for about 2 science/sec.

[PurpleGreen]

unintuitive, as i need to place a chain signal first, and not a signal.
to me a chain signal, should be, well, chained to another, not the other way around.

also right of way is not immediately grasped by many people. we tend to think of tracks as pass/no pass, and not as "oncoming traffic, leaving traffic"

to me its a backwards system. a signal should be two way. and a chain signal should chain off a signal.

chain signals look to the signal before them. so .... i believe you are doing it wrong, misunderstood the concept .

[Elok]

1)
Yep. Belts have lower throughput and far more lag end to end.
And you can't use the belt for other types of shipments.

Lower throughput, yes. And I've always said that rail have higher volume than ore and scale exponentially (not much use to go over 4 rails) while belt grow linearly. That being said, filling +20 blue belt (800 ores per seconds) is a "LOT". Yeah I'm sure there's some base that beats that, but I think we can agree that 800 ops cover +99% players bases.

As for the lag, I don't agree. There's nothing keeping you from adding huge equalizer at the entrance of your base where you make sure you're stable even if one of your mine dry out (same as the train).

Let's put it that way ; look at your big train unloading station. You could replace the train by a belt highway and it will react mostly the same. And for a bonus, you have a pretty great visual of your buffer and you can see more easily when a mining base is starting to dry out.

2)
Rails are omnidirectional.
You can run trains back up the track, and put a factory where the ore patch was without changing anything except the stops. The track will also continue to be used as an alternate route through your base if necessary.

True, but what do you send to your ore mining station that will require a train?

And unless you want to build the whole belt highway by hand, the whole thing will be connected to your drone network so you can send stuff over there quite easily with them. But you need to be careful with branch as drone could take a shortcut and fly over a bitter nest.

3)
No. Why would you even consider this valid for a second?
Using drones is just as easy, but faster and cheaper with rails.
Rails are so extremely cheap that you can even roll along at a decent clip in a train and use your personal roboport to lay long track without needing infrastructure.

Drone will build rail faster, true (8x time faster if you compare 4 rails VS 16 belts). But drone work is background work so it's pretty much meaningful and should not have any effect on your progression. And when you have a base that eat over 800 ops, drone supply usually isn't a problem.

As for cheaper, yes but, as I said in a previous post, even if belt highway are more expensive by 1-2 order of magnitude, it's still negligible compared to the resource you'll get from a ore field (considering a default game).

So, in the end, building belt instead of rails is a lot more work...for drones. But who care about their feelings? Make them work work!

[Frightning]

Well even a "small" post rocket base which only does 600 science per minute needs about 30k iron ore + 30k copper ore per minute these days. So the 12k total from that 8k iron + 4k copper network would be enough throughput for about 2 science/sec.

Yes, demand for raw materials is much higher now thanks to the way science was changed, back then, that was enough for me to get off over 20 rockets while still working on the base in ~110 hrs time, and I wasn't even trying to be fast at it.