Bots versus Belts

[Hannu]

The problem late game with belts is that for massive throughput you need massive belt lanes. And then most often you need to figure out how to balance those massive lanes. Where as bots can inherently self balance if setup correctly and have almost unlimited throughput by just adding more bots and upgrading speed. Ups isn't really an issue for me lategame its figuring out the large space needed for using massive belt systems

Most production can be planned to work without those crazy balancers. Most players get strange reactions if something is imbalanced but in most cases overall throughput is not affected. Of course there are situation in which balancing is needed, like train unloading, but it does not have to be perfect full throughput balancing. Even low throughput systems will balance things during time. It must be another symptom of the same disease which makes people to build absolute ratios.

[nafira]

Maybe some kind of high speed belt that can't be interfered with except at the input / output points.

You mean a train?

But seriously, a single belt is highly optimized. The UPS drops occur when you put splitters everywhere.

Yeah, maybe that's the spark, belts that feed directly into assemblers.

Like that but with trains instead of truck :

[blazespinnaker]

Clever and obviously correct approach - taking inspiration from real world.

In that vein, how about machines which do assembly on the belts themselves without actually removing it from the belt. Quick google searches show it to be pretty common, eg:

https://www.pngfuel.com/free-png/ahuhb

As has been noted, it's the belt -> inserter -> chest / machine lookup which is causing a lot of UPS issues. It could even be done in a way that has different ratio / optimization problems. Like maybe much faster and much higher throughput than regular assembler machine, but some error rates occur and you have to build a quality control mechanism which melts the items down and recovers some raw components.

Honestly, this is the biggest issue I'm running into at this point. UPS bottlneck issues. It's also the supreme and lasting competitive moat factorio has over newcomers to the building video game biz, the hyper genius optimizations WUBE have made to factorio which allows you to build at such scale. No amount of clever features and graphics, in my mind, will ever overcome that. Faster computers will just mean I can create bigger worlds.

In fact, I could quite happily just play in map mode, so little do I care about graphics are other things. It's the scale that is so much fun.

It might be interesting to come up with a meta idea around that as the UPS issue is improved and scale becomes more important.

Maybe after launching the rocket you discover something about the planet and its inhabitants, so you start to research about communicating with the natives. Maybe you discover their planet is about to be hit by a super nova wave or something and you want to save them all by getting them off planet and have to launch billions of rockets. I think it'd make a very nice lore - at first you start out destroying them and then you discover the indigenous species are actually intelligent in their alien way and are self aware.

Or maybe to keep up the PVE theme that so many like, you find out the planet is about to be attacked by killer robots and need to launch a planetary defense in both space and on land. Instead of attacking the natives, you help protect them.

I think it'd be good to make the majority of the alien communication something you can only really do after launching the rocket. Maybe some earlier tech that helps you be diplomatic enough not to pollute and be on relatively peaceful terms, but the real meat of it should come after your first rocket launch.

[blazespinnaker]

The problem late game with belts is that for massive throughput you need massive belt lanes. And then most often you need to figure out how to balance those massive lanes. Where as bots can inherently self balance if setup correctly and have almost unlimited throughput by just adding more bots and upgrading speed. Ups isn't really an issue for me lategame its figuring out the large space needed for using massive belt systems

Most production can be planned to work without those crazy balancers. Most players get strange reactions if something is imbalanced but in most cases overall throughput is not affected. Of course there are situation in which balancing is needed, like train unloading, but it does not have to be perfect full throughput balancing. Even low throughput systems will balance things during time. It must be another symptom of the same disease which makes people to build absolute ratios.

Sorry, what's the math behind that? JIT demands that you optimize perfectly so there isn't anything less/or more than what is required to build. And factorio is all about JIT.

Unless your saying the game can be enjoyed without perfect optimization, in which case I obviously agree. Just random adhoc building is huge fun and should never be completely punished.

But to say that balancers don't matter..

[DaveMcW]

But to say that balancers don't matter..

If you are doing hardcore UPS optimization, you never build balancers. You can even put trains on a timer so they leave without being balanced.

[Rjskeet]

But to say that balancers don't matter..

If you are doing hardcore UPS optimization, you never build balancers. You can even put trains on a timer so they leave without being balanced.

Sadly splitters are extremely ups heavy and making balancers requires splitters. However you could have a hybrid system if belts and bots and use the bots as your balancers

[blazespinnaker]

Ah, well, there are many ways to balance without splitters I think

[jodokus31]

Many people use balancers just to mix like 32 belts to 32 belts. This way the bottlenecks (belt gaps) get distributed evenly.
That does not make very much sense IMO unless you are starving on resources and/or you don't know, which consumer needs what.

if you have a production line, which constantly consumes one belt of iron non-stop, you can simply give it one belt of iron. Why should it be mixed with other belts?

A better use case is take partially filled belts f.e from miners to produce full belts. But i would rather speak of a merger here.

[Premu]

Ah, well, there are many ways to balance without splitters I think

Yes - if you ensure your production lines will produce already balanced output, you don't need to put a balancer afterwards.

For my new copper smeltery I plan to produce 48 blue belts full of copper plates - without using any balancers. Splitters are just used for directly unloading or loading the ore resp. the plates into the train cars.

[Koub]

Yes - if you ensure your production lines will produce already balanced output, you don't need to put a balancer afterwards.

Producing evenly is only half of the solution : one still must draw evenly from both lines of a belt.

[blazespinnaker]

Ah, well, there are many ways to balance without splitters I think

Yes - if you ensure your production lines will produce already balanced output, you don't need to put a balancer afterwards.

For my new copper smeltery I plan to produce 48 blue belts full of copper plates - without using any balancers. Splitters are just used for directly unloading or loading the ore resp. the plates into the train cars.

I've started to avoid any train loading for such things as copper plate because of the inserter lag.

If you're playing railworld, I guess that'd be different, but I try to build everything I can via belts and only load science onto trains. Though even that I could probably get away with by having localized labs. When a particular area dries up, I tear everything down and move onto the next patch.

That said, I'm not entirely sure how badly long belts are impacting memory, but generally my belts don't get that crazy long, and if they do, I skip to an area where the patches are closer. I suspect as my mining efficiency grows and I get further out, it'll become less of an issue.

Also, my computer has 32GB and is generally more CPU constrained (over heating issues)

[astroshak]

Many people use balancers just to mix like 32 belts to 32 belts. This way the bottlenecks (belt gaps) get distributed evenly.
That does not make very much sense IMO unless you are starving on resources and/or you don't know, which consumer needs what.

if you have a production line, which constantly consumes one belt of iron non-stop, you can simply give it one belt of iron. Why should it be mixed with other belts?

A better use case is take partially filled belts f.e from miners to produce full belts. But i would rather speak of a merger here.

One thing I tried, and works most of the time, is a combined pair of mods. Combine loaders or miniloaders with a mod that gives you larger idea, either by merging chests or through a warehouse type of larger box. Obviously if you could filter the slots of the box (a la cargo wagons) you could use them in lieu of belts. The idea here, however, is that you use the mini/loaders to build the load the ore/plates into the best, and remove To continue on down the belts, all nicely balanced.

I’m just not sure if chunk borders affect the smooth operation.

[Premu]

Yes - if you ensure your production lines will produce already balanced output, you don't need to put a balancer afterwards.

Producing evenly is only half of the solution : one still must draw evenly from both lines of a belt.

You're right - in my case I ensure it by unloading the ore from a 8-car train to 16 blue belts (one belt each on one side), and all belts coming from one side feed a furnances which output is put onto another eight blue belts, all going to one 4-car train delivering the plates to the consumers. In this specific cirumstances eight belts will always have the same amount of ore resp. plates on it, so the balancer is not needed.

[Hannu]

Sorry, what's the math behind that? JIT demands that you optimize perfectly so there isn't anything less/or more than what is required to build. And factorio is all about JIT.

Factorio has nothing to do with JIT. In real life everything costs and JIT is a way to optimize costs. In Factorio everything is free and infinite. All costs are rolegaming and very personal opinions is the cost something in Factorio world, player's time or CPU cycles. Or something between them.

Unless your saying the game can be enjoyed without perfect optimization, in which case I obviously agree.

I say very strongly that I enjoy the game without perfect optimization. I enjoy large complexes of seemingly working mechanical machines. I prefer very complex mods which change vanilla's straightforward tree-like production chain to more complex network with loops and side products. I can not understand what is so enjoyable in optimizing simple videogame perfectly. As in real engineering, you get usually 90 % of production with 10 % of time and finalizing last percents explodes costs exponentially and is of course usually avoided in work. I understand it may be fun puzzle in game for someone but not for me.

Just random adhoc building is huge fun and should never be completely punished.

There is a huge gap between random ad hoc and mathematically perfect buildings. I try to get 80-90 % performance.

[blazespinnaker]

Factorio has nothing to do with JIT.

heh heh ok

Factorio is whatever you want it to be.

That said, I'd be curious to see someone try to complete wave defense without optimization

[TheRangerLOL]

That said, I'd be curious to see someone try to complete wave defense without optimization

have you done it yet?

[Qon]

Producing evenly is only half of the solution : one still must draw evenly from both lines of a belt.

Why? If the belt isn't fully consumed and only one side is consumed at first then the thrughput of 2 belt sides isn't needed anyways.

[Koub]

Indeed, but in that case, given enough time, the wagons emptying on these belts where draw is uneven will will take more time to empty, which will in turn lower the throughput on the other lanes (because one can't have 3 wagons leave the stations as soon as they are empty, leaving the remaining wagon to empty at its own pace).
The issue wouldn't be an issue with 1 wagon trains though.

[Premu]

Indeed, but in that case, given enough time, the wagons emptying on these belts where draw is uneven will will take more time to empty, which will in turn lower the throughput on the other lanes (because one can't have 3 wagons leave the stations as soon as they are empty, leaving the remaining wagon to empty at its own pace).
The issue wouldn't be an issue with 1 wagon trains though.

Or you could alternatively set a depart condition if already one wagon is empty. You need to use some circuits to detect this first, though - I did that once exactly this reason.

The disadvantage of that approach is that you send trains back to be loaded which are not completely empty and thus create unnecessary traffic in your train network.

[Koub]

Hence the good reason for balancing not only belts, but also line on the belts