Thesis: Its trains that replace belts

[Lastmerlin]

More food for thought connected to the belt/bot discussion, but hopefully different enough to justify its own thread.

Thesis: At the end, its trains that replace the function of belts (not bots).

Argument:
For almost all bases (except totaly chaotic ones), you can distinguish production lines and the interconnecting network. A production line is a set of assemblers creating the same product from a common input source (e.g. A line of battery plants sharing a belt for copper/iron and a sulfuric acid pipe). The network inbetween is responsible for somehow delivering the inputs and shipping out the outputs.

My main argument is: Most of the challenge and gameplay fun always lies in the interconnecting network. The production lines are mostly simple, and quite repetitive (e.g. building the line of battery plants is quite simple, once you have the inputs ready). Just switch off the additional info (ALT) and try to identify the products of different lines. Thats difficult, because they are so similar. You see two parallel belts, a line of assemblers grabbing from them with fast and long inserters and outputting on a belt on the other side. Could be anything. The difficulty is usually the placement of the production lines and their interconnection (like, "woot I need stone here o_O, how can I get another belt in there and where is the next stone patch"). This is what creates all the funny spaghetti. By the way, this is also, why the main bus approach is both so popular and condemned as boring: It gives a simple strategy for organizing this interconnection fabric.

At the beginning, both the production lines and the interconnection are realized by belts. As the production grows towards megabase the single production lines grow vastly bigger, pushing them further apart. This makes belts as interconnection medium inefficient. There are cases, where they are replaced by bots (see the monolithic base posted by Twinsen), but the usual answer is: Trains. Especially, when going really big, they are the only option. That can be observed in any *huge megabase* lets play series. Its true that bots replace belt concerning the production lines. However, the production lines are not the big challenge anyway, and at this scale they come down to simply plopping down a blueprint. The more interesting part is always connecting, organizing and balancing them. In this regard, belt are replaced by trains, because they are cheaper, faster, offer far more troughput and are even easier to use at this scale (look at these huge balancers, if you need an argument).

Thats it, I hope for a nice discussion.

[joonnnyy]

this is a very interesting though and in reality it would be true, but we have definit and unchangeable circumstances that favor belts over trains in certain areas, so that they complement each other instead of replace

train has advantages in throughput and speed
belt has has advantages in permanent independent movement and size
size because a belt can do a 180 on 2 tiles a train needs ~20 i think, coupled with the fixed size of all buildings that is much smaller than the train would need
permanent independent movement means that whatever is put on the belt moves by itself (backlog from full belt not included) without any additional input
meanwhile trains are objects that do the movement for the item, so without train no movement
loading a train can also not be done anywhere, especially not while other trains are using the same tracks

but belts can never ever reach the troughput and speed of trains
so trains are very good to move stuff far away, which belts cant do good
but they fail at delivering a few items to small row of assambly machines and picking up the result, whereas belts can do that good


meanwhile logistic robots may not have the permanent independent movement of belts, but they make up for it with the extremely granular and "spaceless" transport taht can be scaled upwards to almost infinit
"spaceless" in the meaning of no space needed for the transportpath like rails or belts and no space needed for each entity so they have no collision in the air
the granularity is 4 items which is very usefull for a single assambler or two
coupled with the almost infinit scalability means there is no downside towoeds belts beside the required setup and running cost
and given enough electricity and bots(40 x stacksize/4 x speeddifference) they could easyly replace trains if used properly
all you need to do is create a logisitcs network that is only the desired path with close to no changes of direction so the bots always fly over the roboports and can reload anywhere on the way
but since this is not easy to use, cost more than the entire main factory given even a medium long path and are per definition not intercoactable, trains are used for this purpose

also trains are more ups-frendly than even bots for all but small distance, while belts...

[SQLek]

You could also go full trains, no bots and no belts. https://www.youtube.com/watch?v=8PXAw6wAbrk

Or direct inserting and using cargo cars as warehouses? viewtopic.php?f=194&t=49194#p285360

I'm not expert in this topics. I hardly build any megabase and only few rockets have send. It was fun experimenting with train-bus, but i always tinkered to much with concept to finish factory

[Lastmerlin]

The idea was not to say, that trains will replace belts everywhere. Perhaps, I should have tried to make the point even more more concise.

Second attempt:
Belts are replaced by both trains and belts. For the production lines, bots replace belts. For the interconnections, trains replace belts. In both cases, the new technology is superior in their area of application in each regard: costs, throughput, speed, ease to use (with high troughput). Everybody loves trains. Why all the blame to the bots?

Which of these replacements is more defining in terms of the gameplay style and which challenges to face? What I tried to underline in my main argument: The interconnection network is the more challenging part (especially for veterans). Hence the trains vs belt replacement is the main substitution. The bots vs belt part does happen, but its a secondary effect.

[Lubricus]

I think I understand what you mean. It's the interconnecting/logistic part between different productions of the game that is fun. Belts have that function in normal bases and is often replaced by trains in megabases.
Could this also imply that more powerful belts would make the game more boring and tedious where it actually would trivialize the fun part of the game and shift it more towards building production that is more repetitive?

[Lastmerlin]

Well I don't believe, there is a risk in buffing belts. Trains are superior to belts for long range troughput scenarios by a factor, that definitely exceeds the estimated factor vs bots for short range throughput. I would guess a factor of 40. It depends on the stacksize of the item, but for plates I did some tests and got over 100K plates per minute troughput, which is equal to 40 blue belts (for a price of a fraction of one blue belt). Moreover, using belts for massive troughput is not trivial, but a headache. Just imagine the space requirements of 40 parallel belts and look at those huge balancers...

[olafthecat]

Belts NEED to be changed in some way, whether being nerfed or buffed.
They are just too unrealistic at the moment.

[Optera]

I fully agree, trains are THE replacement for main bus belts for large bases.
Not only do trains offer more throughput and simpler routing than a belt main bus, they also allow for a completely distributed approach (towns) which is infinitly simpler to expand than a main bus.

[pgriss]

How would you rate the difficulty of building a late-game base using

1) Belts + trains, no robots
2) Robots + trains, no belts
3) Trains only, no belts or robots

?

I'd say if your thesis were true, building a base with trains only should be no more difficult than building a base with belts + trains.

In reality, however, "belts + trains, no robots" and "robots + trains, no belts" are both significantly easier than just trains. This to me shows that trains don't replace either belts or robots.

Trains do replace the main bus. Nobody minds because trains are a lot more fun than the main bus and eliminating the main bus still leaves room to use belts elsewhere.

[Lastmerlin]

If you want a complete consideration, you have to distinguish 6 cases:

A: Belts only: Almost impossible, no good long range transport method. You are capped by what ressources you have nearby and have no good means of dividing the factory in sub-factories. You could run dozens of very long parallel belts instead of trains. Looks just very ugly and kills UPS in 0.15 (way better in 0.16). Limit: < 500SPM in 0.15, better in 0.16 (perhaps up to 2k SPM?)
B: Bots only: Impossible, no good long range transport method. Using robots for long range transport with long lines of roboports looks completely insane to me. You have to cover the entire area between your subfactories with roboports, otherwise the robots will fly over the no-mans-land and get very slow. Insane energy costs and also probably deadly to UPS. Limit: <500SPM.
C: Trains only: Very difficult, but doable (and really cool). You need direct insertion from/into train everywhere. There are examples on reddit. You need a few more machines, because beacon coverage is not optimal. But I believe with very intelligent train network, this is even really UPS efficient. Limit: I guess 5K SPM+
AB: Belt+Bots, no trains. Still almost impossible. You can use huge belt arrays to connect subfactories and keep bots in disconnected networks. Ugly, but works (and even UPS-efficient in 0.16). You can use bots for you subfactories, but thats no big win, if you have to put the stuff on belts for longer range transport anyway. Limit: Same as belts only (awful in 0.15, way better in 0.16)
AC: Belts + Trains, no bots: Difficult. Main difficulty is balancing and loading/unloading. And of course the train network/production control. The production lines are not that difficult really (I have belt/beacon blueprints for most large-scale products and bet most mega-base players have). Limit: 1k SPM in 0.15 (due to UPS), better in 0.16 (perhaps 3k SPM?)
BC: Bots + Trains: Difficult. Main difficulty is the train network/production control. Production lines are simple (although fine-tuning can be tricky). Limit: 5K SPM+ (perhaps 10k SPM on modern machine?).

Why no rating below difficult? Anybody says that bots make the game trivial. My experience: Building a mega-base is among the most difficults tasks. Deathworld, no spoon, lazy bastard all felt way easier and took less time.
Why no rating difference between AC and BC? For the same final production BC (bots/trains) is slightly easier. But a real mega-base builder strives for the limit. My BC (bot/train) basis reached four times the production as my AC(belt/train) basis did before UPS dropped. Hence, the (bot/train) base provided more challenge.

Final remark: The thesis was provocative and a bit too simply stated. I clarified my idea later on:

Belts are replaced by both trains and belts. For the production lines, bots replace belts. For the interconnections, trains replace belts. In both cases, the new technology is superior in their area of application in each regard: costs, throughput, speed, ease to use (with high troughput). The interconnection network is the more challenging part (especially for veterans). Hence the trains vs belt replacement is the main substitution.

Edit:
Someone posted this base on reddit today and it _really_ deserves some referencing:
https://imgur.com/a/SXfs1
This is belt+trains. What it shows (at least to my eyes): The production lines itself do not look terribly difficult, although a bit more interesting than bot layouts. Size also makes no real difference here, you just blueprint 20 copies instead of 5. What is absolutely crazy and stunning is the fabric inbetween. Here it is a crazy train network and a huge belt tangle on top. Very impressive and the real brain smelter of this base. It also shows, that belts are indeed a viable option for longer range transport in 0.16. Seems, that my estimate of about 3k SPM as limit was quite correct. The devs really did an excellent job at optimizing here.

[HurkWurk]

actually bots only is easier than you think, because you just make separate networks and line both sides with requester > loader > passive supplier chests across the gap, so that as things are brought to the edge of 1 roboport area, they are transferred to the next so that your robots never actually fly long distance. IE its a giant relay race instead.

[Lastmerlin]

Ok, its apparently not impossible, but just very inefficient and quite stupid.
If I wanted to transport some bulk goods over 1000 tiles, I am pretty sure that 20 blocks of logistic network are by far the most costly option, both by energy/ressource but also UPS. Belts and trains are way better and might be actually tied regarding UPS in 0.16 .

[HurkWurk]

the nice thing about doing it via relay races though, is that until you get to the edge of your main factory area, you can actually scale massively, and not have issues with depleting locally while waiting for the next train to load. (since there is no train load/unload cycle)

but yes, definitely the most costly.