Train unload poll

[ChurchOrganist]

We seem to get a lot of belt train unloading systems posted in "Share Your Creations".

But I've been wondering how many of you use bot systems instead like I do.

If you wish to alaborate on your vote, feel free to do so in the comments.

Poll will close in 3 months.

Go!

[Frightning]

Bot based train stations are superior in almost every way except cost. It's less of an advantage on the loading end, since you don't often need extreme throughput (exception: huge ore deposits), so belts are often sufficient. But in both cases, they are simpler to setup up (no need to worry about balancing the (un)-loading because bots do that automatically). On the unloading end, they enable one to have 'universal' unloading stations which can handle unloading of a variety of materials at the same station without any sorting issues (much harder to achieve with belts).

[Hammerchief]

Bots may be better, but they are boring. Trying to do everything only with belts is definitely more enjoyable for me than just place roboport and chests.

[ChurchOrganist]

Bot based train stations are superior in almost every way except cost. It's less of an advantage on the loading end, since you don't often need extreme throughput (exception: huge ore deposits), so belts are often sufficient.

Currently I load ore etc with belts as it saves on setting up a logistic network at the outpost, and is self regulating in a way that using provider chests on each mining drill is not. Once the belt is full mining stops, so that pollution is controlled to a degree. I suppose you could use passive providers on the mining drills, but they hold an incredible amount of ore unless you limit them.

Unloading is another matter - using stack inserters to unload into active providers means the ore gets transferred to smelters very quickly.

One possibility to make things more efficient might be to try smelting at the ore outpost in use trains to bring in plates to the main factory. I might try that on the game I started at the weekend.

[Yehn]

Bot based train stations are superior in almost every way except cost. It's less of an advantage on the loading end, since you don't often need extreme throughput (exception: huge ore deposits), so belts are often sufficient. But in both cases, they are simpler to setup up (no need to worry about balancing the (un)-loading because bots do that automatically). On the unloading end, they enable one to have 'universal' unloading stations which can handle unloading of a variety of materials at the same station without any sorting issues (much harder to achieve with belts).

I use belts for loading, bots for unloading.

Even with a huge ore deposit and reasonably frequently trains, express belts will keep the loading chests full (with a proper setup... 2 simple BPs is all I need to balance it).

But since unloading stations are usually shared (at least for me...), I really need the station cleared ASAP. Bots can't be beat for that.

[OdinYggd]

I usually use a hybrid of the two systems:

Bulk cargo like ore, oil, and plate is loaded and unloaded by belt. Getting a uniform draw across all of the boxes is tricky, but between combinator magic and belt splitters it can be done.

Incidential cargo like repair kits, bullets, and spare bots, I usually just use a bot-based solution to handle sorting and storage with the inserters simply trying to maintain the network counts in a specified range. This keeps the layout simpler.

And then there is coal, which is handled both ways. All of my trains carry a couple of stacks of coal in the first car as a reserve fuel for the engine, this is transferred by logi bot from the unloader to the requester and inserter that puts it into the locomotive. Like so the trains never ever run low on fuel, because whenever they stop more is being added.

But a remote coal mining site will almost certainly use my ore pattern of belts onto and off of the train, because it is a bulk cargo and belts are superior to bots when it comes to moving bulk goods.

[SpeedDaemon]

Like Odin, my setups usually depend on the circumstances.

0.13 came with one huge change as far as some unloading stations go, though. That being that we can now tell a train to "wait till empty", so there's no point in having buffers. My ore gets unloaded from the train directly to belts, set up so that as soon as it's empty, another full one can pull in behind it. Stack inserters and judicious use of splitters can easily keep things balanced.

[Frightning]

I usually use a hybrid of the two systems:

Bulk cargo like ore, oil, and plate is loaded and unloaded by belt. Getting a uniform draw across all of the boxes is tricky, but between combinator magic and belt splitters it can be done.

Incidential cargo like repair kits, bullets, and spare bots, I usually just use a bot-based solution to handle sorting and storage with the inserters simply trying to maintain the network counts in a specified range. This keeps the layout simpler.

And then there is coal, which is handled both ways. All of my trains carry a couple of stacks of coal in the first car as a reserve fuel for the engine, this is transferred by logi bot from the unloader to the requester and inserter that puts it into the locomotive. Like so the trains never ever run low on fuel, because whenever they stop more is being added.

But a remote coal mining site will almost certainly use my ore pattern of belts onto and off of the train, because it is a bulk cargo and belts are superior to bots when it comes to moving bulk goods.

Ya know, I've heard a lot of people say this, but in my experience that really isn't true. Yes, the belts (especially lower tier belts!) are cheaper than bots. But a decent sized bot system has enormous throughput potential over shorter distances (especially with cargo and speed upgrades). I have a base that uses only logistic bots to move everything, which is consuming ore at a rate of about 2250 iron and 2250 copper a minute, and since getting the last of the bot upgrades I rarely ever need more than 900 of the 2900 bots I have in the system. Mind you, it's not just moving ore to the smelters, but EVERYTHING needing to be moved in this base that is being moved by ~900 bots or less.

[Aatch]

I do both, depends on the situation. I generally try to work with belts, but often it's not worth the contortions. I normally use logistics bots for refuelling, as it means I can have refuelling at lots of different stations without needing to figure out how to run lines to it.

[OkariDraconis]

I've personally done both Bots and Belts for unloading and Loading..

Currently though I've found belts w/ stack insertions to fill my needs. The main difference is now I have 3 stops at my main base, (CopperPlates, IronPlates, Oil).
Additionally I a Standalone Copper Smelting base, and Iron Smelting Base. Essentially, I route all Ore trains to it, then have a separate station for pickup (of plates).
Oil trains are 2 cars only, one car for Oil, the other for Empties.

By having the (Electric) Smelting as a stand alone base, it allows me easy expansion of smelting operations without effecting any other area. Additionally, Setting up new ore mining bases is a snap. Were I only have to worry about getting the ore mined, and loaded to train, then just route it to smelting. My smelting operation is double buffered on both the Intake and outake. This way I can have more ore than it can handle going into it, and additionally, if the demand for plates is low, the smelters don't get backed up, and I get a nice resereve delivered by train when needed.

On a different side note, I played with bobs mod, and had a similar distributed base, eventually I just used bots for everything since those tier4 bots are ridiculously fast moving and cheap on energy.

[TruePikachu]

(These statements apply to 0.12; I haven't really gotten to the relevant point in 0.13 yet, due to stuff.)

I generally like having my entire base inside logistics coverage, so I use belts for train unloading (though I tend to use bots for loco refueling).
On the other hand, I keep remote outposts inside construction bot coverage (for automatic repairs), so it might be feasable to use bots for loading.

[Hannu]

Bots are probably technically better way, especially in huge factories, if you want just best throughput with minimum planning and building work. However, I feel that planning and building of (relatively) high capacity belt systems is one of the most interesting things in Factorio. I also like that complex belt systems are more aesthetic than insane number of robots on zig-zag-routes. Therefore I restrict my robot using to low volume transports and use always belts in high volume ore handling.

[ssilk]

I think the biggest advantage of bot based unload is, that it can handle a special case: A multipurpose-unload-station.

When is this useful?

Assumend you have some outposts and the outposts produces copper, iron and stone each (in different amounts).
Now instead of having one train for each resource on three train station for each resource to load (and three to unload), you can have three trains going to one train station. Any item-type is loaded into the trains and on the multipurpose-unload-station it is unloaded and then the bots sort the stuff out into the different directions...

Of course you need to control production then somehow, cause you cannot use locked-buffer-production for this. But that is a different subject.

[ChurchOrganist]

Of course you need to control production then somehow, cause you cannot use locked-buffer-production for this.

Yes indeed!

I've just got my first train going in my new RSO map that I'm playing which delivers iron ore to the main base.

I found some oil patches near the ore outpost so spent some time setting up pumpjacks for those and adding a rail spur to pick up barrels.

On returning to the main base about 30 mins later I discovered that my bots had squirrelled away 56k of iron ore into main storage!

The original intention was to make this a multi-ore delivery point, but I need to work out how sensible storage levels can be maintained.

[MeduSalem]

Bots all the way...

Makes designing a Trainstation a lot easier and one can cram in a lot more parallel stations in the same space because one only needs 2 tiles (Inserter + Chest) on each side.

Ontop of that what ssilk said... they basically do the sorting for you... so no belt/splitter spaghetti or other contraptions.


Currently I'm doing it so that my main stations are surrounding a central mass storage (4 stations on each side of the mass storage... so a total of 16 unloading stations)... so the Inserters directly unload from the trains into Active Provider chests from where bots pick the stuff up and store it into a huge array of Storage Chests. The Inserters work on condition and only unload if the mass storage isn't overstocked with a certain item. And I use the contents of the Logistic Network to send signals to the outposts to determine which resource should get loaded into a train and if a train is allowed to depart in the first place. (All main stations have the same name, all outposts have the same name)

My factory in general is bot-optimized... the various factory departments are also surrounding this central mass storage (outside the trainstations of course, the train stations are closest to the central storage)... so input resources for a recipe are picked up from the central storage... the output products are stored back to the central storage.

[dragontamer5788]

I think the biggest advantage of bot based unload is, that it can handle a special case: A multipurpose-unload-station.

I've accomplished this (WITH backpressure) using combinators and filter inserters. Its a damn shame that stack-filter-inserters only have one filter so I must use purple filter-inserters.

The idea:

1. Use the circuit network to "set filters" of the filter inserters. Only offload cargo that is "in demand".
2. The trains know what items they have and should have filters set on the trains. Train should leave if (Iron Ore is not demanded) in the circuit network. I use negative numbers to indicate demand... so Train leave immediately if (Iron Ore < 0).
3. I can use stack-filter inserters to sort out the stuff in different directions.

Bam, solved all issues using belts only! (although a bots-solution is equally easy. Just hook the combinator network to the logistics network and count items available) The only problem is the slower-throughput from using purple filter inserters instead of white-stack-inserters for offloading. But being able to filter-out 5-different items is a necessity with this kind of setup. Another "problem" is that train-traffic will be higher than necessary. Trains will constantly be zipping around everywhere when backpressure is high (Oh, this station is full. Going back around my routine... at Iron Ore, the train will then be "full" already, so it leaves immediately. Etc. etc.)

That problem can probably be solved with a "blocking" train station waiting-bay. Trains only leave the waiting bay if the main-station indicates demand. (ie: Pre-station has Wait forever until Iron Ore < 0). Haven't really tried that yet... but sounds theoretically safe.

I dunno if there are other schemes, but this one definitely works, albeit a bit slower. 12-filter inserters is still kinda fast though.

[ssilk]

Without seeing it, I go on thin ice, but I think - also without seeing such a station - that such a belt based station cannot compete with the throughput compared to a bot based. I'm not saying that bot based station has always better throughput, but for this case (multipurpose) it will have. And of course: Every exception has it's exception.

What I also want to say is, that I don't understand this kind of grimm.

Belt transport vs. Bots. Bots vs. Trains. Trains vs. Belts.

It remembers me to the "fight of programming languages": C vs. Java. Java vs. PHP. PHP vs. Python. Python vs. Perl. Just to name some of those; search on the internet and you'll find.

So, logically, looking on it from this side, this kind of poll makes no sense. Cause the "right way" is in my opinion to look at something and then decide what's the best tool, the best way, the best method and so on, to achieve the target. And not: "I build my stations always with belts. Howgh!"

[Hannu]

What I also want to say is, that I don't understand this kind of grimm.

Belt transport vs. Bots. Bots vs. Trains. Trains vs. Belts.

I do not like such competitive attitude too, but this kind of polls give interesting information about other player's playstyles. It also helps to understand many game development decisions which are against my personal style. In my opinion these have been quite correct discussions (or moderators have done their job very well and I have never seen problems).

[ChurchOrganist]

What I also want to say is, that I don't understand this kind of grimm.

Belt transport vs. Bots. Bots vs. Trains. Trains vs. Belts.

The poll was never intended to be a Bots v Belts or anything else war.

As I already said further up I use both - belts for loading trains and bots for unloading trains - at the moment - that may change depending on game circumstances.

I was just interested in the split of users between the two systems, but more importantly I was hoping that in describing their systems I would learn more about train loading/unloading systems and thus improve my game.

I'm pleased to say that this is exactly what has happened.

The combinator stuff described looks interesting, but is an area of the game I haven't properly explored yet. I'm currently thinking that combinators may be the way to control whether the inserters load the active provider chests from the train or not depending on existing ore stocks, but I need to give some serious thought about how I can achieve this before I try to implement it.

[Tricorius]

It remembers me to the "fight of programming languages": C vs. Java. Java vs. PHP. PHP vs. Python. Python vs. Perl. Just to name some of those; search on the internet and you'll find.

Java and PHP are obviously horrible. One must bow to our new Ruby overlords...

On topic, however...

I generally use a combination of both. I usually have *very* large bases. I have rather large rail yards (I consider trains essentially an extension to belts) with at *least* one station of each type of bulk input (iron ore, copper ore, stone, coal, oil). Each of these stations has a waiting bay of at least three train lengths (I generally run 1-4s, but you could easily change that to 1-4-1s--loops are not evil if you know how to run them--but that's another argument for a more civilized time).

For instance, my current base has two each of iron ore and copper ore unloading stations, one stone, one coal, and one oil (barrels). Each station has a waiting bay of three trains for a total of 28 train lengths (including the trains in the stations). Granted this is complete overkill for oil, but that's beside the point.

In addition, I have several engineering trains with load outs meant for building / resupplying remote outposts (I like to keep my outposts under small, isolated roboport networks and stay on projectile turrets as long as possible, so these need to be resupplied).

I also use a few miscellaneous trains (such as a spam train that contains all the junk I want recycled back into the base and obselete, burnable items to be tossed into the boiler lines).

As you can imagine, this takes up quite a lot of space, just for the rail yard. In order to keep bots more localized (even though I have full robot base coverage I don't like them flying across the entire base every trip) I run the bulk goods in with belts through the base to the main smelting operation (ores and stone) / power systems (for coal). The exception to this is oil. Bots (amazingly) can transport massive amounts of oil barrels quite efficiently. So I just let them handle oil (my current base has a stockpile of 5.5k full oil barrels in storage since I have my circuit network set to only use them if the main pumps aren't able to maintain 80% oil capacity).

Locomotive refueling is done via bots as soon as possible (and I usually have at least early bots set up before trains).

The engineering trains are handled completely with bots (as they are mostly sporadic, the only regularly running engineering train is the resupplier...even then it's perfectly served by bots since it's more compact and it really doesn't process huge volume).

I'm honestly not too concerned with "efficiency" and "cost". The math would probably prove that this may not be the most efficient way to do it. But I've never run out of resources in Factorio (at least not without it being my fault for not prepping a new outpost soon enough), so I don't care about the relative "cost" of bots or belts or trains. And at some point in time your base is going to be large enough to meet your goals, so it will eventually back up. Therefore "efficiency" isn't really that big of a concern in my mind either. (The only time I've really been concerned about that is when you have an intense shortage of a given resource in your starting area, primarily oil or iron. And at this point I skimp a bit on production and streamline to military so I can expand to a resource outpost when necessary. I really still don't think bots vs belts matters that much at that point.)