Anyone standardizing on very long trains?

[T-A-R]

That looks very clever. I could share some pictures as well.

Honor where honor is due, the designs are heavily inspired by Stevetrov's direct insertion builds. The labs are botfed. The mall on the right is supplied by 48 wagon trains:D

Rail Crafting.png (1.21 MiB) Viewed 6285 times

Paving never finished:p

The new base (WIP) has a isolated flask train loop. Silo's are belt fed. Intermediates will be fed by rails from outside, crafted in outposts, closer to the large ore patches.

Vr21.png (1.18 MiB) Viewed 6285 times

I played other maps aswel to keep variaty in gameplay. And currently am "enjoying" a break aswel.

And to finish off a Mining spiral.

8-48beaconmine.png (162.65 KiB) Viewed 6285 times

Very large patches required (or you need even more stops).

[blazespinnaker]

reduce crossings

yeah, semaphores are evil. undermined by our own cleverness.

[blazespinnaker]

practically required for this:

As fluid wagons don't work on stops where one of the mid cars is on a curved track, you need to do a small tweak.

Someone could create a mod, but as I don't do multiplayer much lately, I just tweaked this:

10 second change in C:\Program Files\Factorio\data\base\prototypes\entity\entities.lua

fluid_wagon_connector_alignment_tolerance = 1.0, (was some fraction before)

It'll probably break for MP I suspect.

I'm sure folks are already doing this, but I've also added a station build wagon. you can set the squares on the wagons to only allow certain items (probably do this with logic as well), so one of the cars always has enough bots, miners, etc to fully build any of my blueprints. As long as I can get a rail there, I can just drop a robotport and paste the station. When the long train arrives, it pulls all the necessary materials from the build car and quickly builds the station. Cleaning up after the base is done is just the reverse, pull everything from the yellow logistics chest to reload everything back on another station car. Makes setting up new stations an absolute breeze. Admittedly, this is less an issue as you grow your infinite miner bonus.

The only minor issue now is laying down rail. I was gridding the entire map with roboports and radar before and it worked amazingly well for a lot of things, but that was a little tedious and I suspect very non fps friendly. Also, not researching logistics so there's that.

You can paste down rail lengths in map mode, so you don't have to draw them out, but you still have to run the length in someway to create the path for your trains.

Fluid wagons are more preferable for sure now. No more dealing with the hassle of laying down pipe.

[wobbycarly]

The only minor issue now is laying down rail. I was gridding the entire map with roboports and radar before and it worked amazingly well for a lot of things, but that was a little tedious and I suspect very non fps friendly. Also, not researching logistics so there's that.

What works for me is setting up a few spidertrons with nothing but roboports/batteries/reactors (no legs) in the grid. Get them to play follow-the-leader with a trunk full of rails/signals/etc. Ensure logistics while moving is enabled and then let them stroll around. It's not quick, but can be very effective as you can do other things while they're off laying tracks.

(I use Spidertron MKIII from Spiderton Extended mod to get larger trunk and grid.)

[blazespinnaker]

Yeah, that's a good idea, will try that for sure.

The other issue I've run into long trains, and I suspect this is why wube discourages it, is that inserter->train tracking gets very random at a certain point.

[mmmPI]

"Longer trains are the most important advance in achieving economies of scale in the past quarter century.”

That's a shame i can't read the original article, they ask me to register.

I was curious to see where in the world did the analysis took place. I anticipated places like Australia, USA, or Russia, vast countries where you have vast area with low population density, and a business around raw material being exported. ( vs places like europe or japan, densily populated, low raw material export.)

Also i was curious on whose perspective was developped, by that i mean from the eyes of a train company, you reason differently than in the eyes of an official appointed to plan urbanism/developpment in region, state, oblast, country or even internationnal transport, like the silk road.

I've learn a few things from years spent doing extensive research on city-building game, if you focus toward realism, you need to understand why things look this way, what is it that you see on googlemap, what it's purpose.

Things that comes to mind are the history of how things were built: in England, Germany and France, who were amongst the first to build a "national train network", were also empire with different level of centralization and free entrepreneurship?, like fully centralized toward Paris, politicaly and physically ( the rails! ) with the state planning to build a military oriented network of troop transportation in France. While in England the "transporting troops" to the border didn't play the same role, due to it being an island surrounded by lots of their own military boats and having started with trains way earlier. Germany wasn't centralized, and trains were built in different states each of them looking like a knot of a web in a map. In those 1800 1900, USA started last for trains, but having such vast territories ended up with way more miles of rails and even more in kilometers, the epoca of the goldrush, and civil war, land given to various private company by congress, all those are just the inital conditions for the current networks.

If you (only) consider the past quarter century, you are looking at a stock graph of the last week for a lifetime investment !

trains are serious business ! haha not so seriously, the networks are built with a very long life expectancy, even if you redo the rails, you rarely modify the trajectory after everyone has built around it right ?, real life like factorio , or i thought so.

I really enjoyed seing the mapview pictures, from T-A-R, i guess the visible stations are not made to be moved often hehe but from blazespinnaker, ( on the one loop thingy) it offers a picture of point in time of a semi-permanent network, it shows a way to manage over time the fact that the ressources input location vary ( for mining ressources not oil is that where you build your assembly line ? ). The unloading station are designed to be permanent, and you strealimned the process of making new station for such long train every time a new mining patch deplete. Also you only have a single lane of track, not like 4, not even 2, just 1 making it even easier to redraw the rail.

I think both mapview pictures are very typical examples of factory planned in-depth to make the better use of long train. ( the spiral miner , avoiding crossing as much as possible) I think the similarities with real life become less strong, but can be extended to freight boats, or passenger planes.

Wether it is for train, boats or plane, the trend to make them bigger also goes with the trend of reducing the amount of stations, harbour or airport, that can host them. because each of them is a bigger and bigger investment as the vehicule is bigger, each of them generating more traffic, covering, on the territories a larger portion of the map, and of the person that needs transportation and goods. Then unlike boats and planes, trains requires tracks layout, and crossing that are not just arbitrary channel and those also needs costly adaptation if/when the vehicule change.

This i see as the main incentive to "standardize". It is thought from the point of view of what would be the logistic manager person, making efficient design considering previous points as core philosophy. It is an economy mostly from the perspective of the network manager.

In real life it is a constant struggle for local official who would want to be connected with infrastructures, but carefully since it would devaluate some properties, which can cost them popularity, but also increase the value of an area, with cheaper freight access, and more activity available in a X-hour radius. The trend of "less numerous but bigger", also called "economy of scale" in some context increases the stakes for each station, each of them generating more jobs/activity, but the opportunities are less and less numerous.

To answer your question more directly, i don't standardize on very long train, the total opposite, i mainly do short trains, and like to have some quirks and weird trains, of different size, or loco placement, schedule and things. I like the way it looks and feel when i play in those maps. What you call nightmare i call beauty, the grid thing has that organic feel of something that grew from a standard DNA/blueprint, but then locally adapted here and there to the reality with as stacker here, a bunch of spagetthi there, some regular assembly line in both direction " where it fit", no wonder why you found the big loop boring haha

Also i leaned toward that due to the annoyance i felt when having to move very long train station. It seemed to me easier to learn how to deal with smaller train, rather than to learn to better plan for long train. It feel to me more versatile, which is good when you can't make up your mind and always want to leave some room for things just in case.

I'm also curious to see if other people are standardizing on very long trains ? It's enlightning to see what doing so leads you to design afterwards. It offers good inspiration to plan for a map without starting your reflexion from scratch.

[blazespinnaker]

Yep, very boring for sure. It’s sort of like taking out biters with logistics artillery on islands.

Extremely simple and efficient with no need to babysit, but that is just it - not much to do. No need for any of the toys.

I guess the only fun thing is that it depletes patches quickly due to the throughput, forcing you deep into biter territory.

You end spending a surprising (at least to me) amount of time laying down new track and setting up new mining ops.

Another game play advantage might be that you don’t really have to plan much ahead of time. You can just organically grow your route and your trains. They can all be of different lengths as well, though that is less efficient.

[blazespinnaker]

Btw, stackers are evil, imho. If a train isn’t loading, unloading or traveling to a place to do so, it’s not really being very productive. Best and really only place to buffer is inside an assembly / furnace. Even then, you want to keep buffers as small as possible.

Factorio is all about JIT. Well, optimal factorio.

[astroshak]

A small (to start) stacker is not a bad thing. Is it better to have the ore waiting at a loading station? Or is it better to have the ore already loaded on a train waiting for a spot to unload? I prefer both.

As distance between ore patch and factory increases (either between ore patch and smelter, or between ore patch with smelter, and rest of the factory) you need more trains to maintain the same trains per minute at the offload stations. Thus, the (increasing) need for someplace for those trains to park before offloading their goods : the stacker.

[Khagan]

Btw, stackers are evil, imho. If a train isn’t loading, unloading or traveling to a place to do so, it’s not really being very productive. Best and really only place to buffer is inside an assembly / furnace. Even then, you want to keep buffers as small as possible.

Factorio is all about JIT. Well, optimal factorio.

I absolutely agree with the sentiment in favour of JIT.

But I think a certain amount of stacking is unavoidable in practice. If a smelter eats a train-load of ore every few seconds, it needs those trains at regular intervals; but they are coming from mines at varying distances with varying production rates. At least a short queue is needed to average out the fluctuations in arrival times. (And while a stack of several trains might look like a large buffer, it may only be a minute or so of actual consumption.)

[blazespinnaker]

I am not seeing it, but maybe it’s because I am playing vanilla. Perhaps with megapatches things are different as you have fewer stops.

For me, the patch loading stops is my “stacker”. Also, every train always arrived max loaded. I keep adding patches to the route until I can assert this. I don’t remove patches until they are fully drained, cause it’s a hassle adding so many patches. (I do miss gridding the entire map with roboports. It turned factorio into creative mod.)

The biggest downside really with long trains is the fact that the entire train needs to leave the area before the next train can start unloading.

I am a bit lazy and don’t do this to the degree I could, but well placed signals significantly reduce the gaps between leaving trains and arriving trains.

So of course, small parallel trains arriving will theoretically beat out thruput of longer trains, but the effort required for guaranteed, resilient thruput is very obscene and likely not UPS efficient.

That said, with a bit of planning and added complexity you could probably make it so long trains could start unloading before the previous one has fully left. I am skeptical the juice is worth the squeeze tho

[astroshak]

Usually I put the inserters on the same side of the track that the signals would be on, but if you are careful you would put a bunch of Rail Signals in between cargo wagon spaces on the rail, allowing for the next train to move up sooner when the old one is leaving. Not sure it is worthwhile, have not tried.

[mrvn]

https://www.wsj.com/articles/why-railro ... %20traffic.

"Longer trains are the most important advance in achieving economies of scale in the past quarter century.”

I have been thinking lately that short trains just suck, period, and would like to start using very long trains. I played around in editor mode, and there doesn't seem to be a limit.

Companies have plenty of reasons to keep adding train cars. Long trains save on fuel and crews, reducing the cost of rail transportation.

I don't quite agree. Well IRL, sure, I guess the journalist who wrote this article knows his stuff at least a little.

But in Factorio, we don't really care about train fuel, we don't have any crew, and there is no money to save. All we're left with is inconvenience of having to design a rail network with huge intersections and huge train stops. There may be advantages to long trains in Factorio (for example higher throughput for a given cost of UPS), but I lack the time, patience and know-how to compare several equivalent train systems (with different train lengths) on that respect.

Lastly, I can't discuss on the cool factor. I guess there is something to brag about "my trains are longer than yours", but honestly, that's the kind of things I don't really care.

The thing about RL trains is that once they roll they keep rolling. A train twice as long and twice as heavy doesn't need twice the fuel once it's rolling. Plus less personal. So longer trains are cheaper. I think it's similar in factorio but a longer train needs a lot of time to get up to speed. Which means you need really long distances and no red signals for it to become more economical. Most people increase the number of locomotives to get the acceleration back up to something usable on shorter distances. Train fuel isn't really such an expense to save every erg of it. So practically in factorio longer trains are better because they have more throughpout.

In real life a 200 wagon train isn't unloaded all at once in a 200 wagons long unloading station. Instead trains are split up or driven through an unloading station one bit at a time. There are mods for splitting and merging trains but no such thing in vanilla. But you can unload a train a bit at a time. Just have multiple train stops at set distances, e.g. 4 wagons apart, and then unload 4 wagons at a time. Unfortunately you can't use curves in this setup with fluid wagons as the pumps won't align anymore once the train passes a curve. So your station footprint will be twice the train length basically, with a small section of unloading stuff in the middle.

Personally I started a new game with the Nullius mod recently using the ribon world for a change. The map is only 128 tiles high, which allows for 2 tracks each at the top and bottom and station going top to bottom with LLCCCCCCCCC>L>L trains. That fits exactly in 128 tiles high. Anything longer I would have to build horizontal or snake around curves.


As for your idea of collecting ore from ore patches in really long trains: I think it's a bad idea. Building such large train stations at every ore patch is tedious and a patch doesn't produce that much ore to need overly long trains. You should rather build one long train station in the middle of a bunch of ore patches and connect it to you base. Then have a bunch of separate rails go to surrounding ore patches and use smaller trains to collect the ore for loading the long trains. You can add surrounding ore patches till the traffic becomes too much for the small trains. And then you build another long train station somewhere else far away and do the same again.

You can also put your ore smelter at the long train station. So the small trains collect ore and the long trains ship plates. Decentralize your factory.

[blazespinnaker]

DI trains are required for UPS. Think about the CPU efficiencies gained, especially at high miner bonuses.

Tedious though. Steel, for example, takes up 1/10 the wagons as raw ore.

Too bad there's no particular good analogy to real life though. Leaky abstracts abound.

That said, your decentralized approach is something I've often thought of. It works sometimes I guess. That map I'm on right now, Copper is a little sparse. Not sure if they've changed the layout logic.

This was discussed in a previous thread, let me find it... viewtopic.php?f=5&t=90401

This dude has some very sweet stuff .. https://factoriobox.1au.us/map/view/d3c ... 3712/0/233

10K spm

[user]stevetrov[/user]

I asked this:

Also, using just belts could probably cut down on your inserter/chest/logic usage by 1/2 or even more. Do DI trains really help that much? Or is it just more fun? Cause that map looked like a lot of fun (ignoring the solar, at least, I really dislike solar)

You can cut down the number of inserters but those inserters have to work for about twice as hard (it takes twice as long to pick up from belts than chests)

It'd be good to do a quick clone tool test to see if this still holds. It makes sense though, just intuitively

My one complaint would probably be the bot usage. I use bots for certain things unrelated to the factory (supplying for base build outs, artillery, fuel), but I dislike using them for the factory itself. It may be more UPS efficient, but it ruins the metaphor for me. Maybe for high value stuff like science or rocket party ingredients, but anything less it just feels wrong.

[mrvn]

Belts got optimized so any past arguments of belts vs. bots is probably inaccurate. You should build both and compare fresh.

As for inserters picking from belts vs. chests. A stack inserter can pick up 12 items from a chest in one check. On a belt every item gets picked up individually and the inserter has to chase the item as well. So you really don't want stack inserters picking from belts and inserting into the train.

A while back there was a discussion about doing direct train insertion. Going from the train directly into e.g. furnaces and from the furnaces directly into the train on the other side. Gets a bit harder with recipes with multiple items in or out. You have to load your trains count perfect then or they can get stuck. I can't remember though if they compared the UPS with bot and belt designs though.

[blazespinnaker]

I think the idea above is simply to replace belts with trains, so you don't need to do furnace/assember DI to furnace assembler. Rather, DI -> trains -> furnace -> trains -> assember -> trains -> ..

It'd be cool if belts were faster than bots, but it wouldn't really change anything for me. Factorio fun is subjective for sure

[blazespinnaker]

Hmm, here's a simple model for DI trains. Doesn't strictly need long trains, but I imagine you'd want them to be as long as possible in order to fully cover a patch.

Basically, fill each wagon with roughly enough raw ore for 15 SPM or so. Do that by rotating the train around patches doing DI. So each wagon has around 35% copper/ 45% iron/ 8% stone/12%coal (something like that, I believe).

Once the train is filled, it goes through a processing factory where each stop removes the raw/intermediates from the wagon and replaces it with intermediates to craft the 15 SPM. Clever uses of assembler to assember is nice, but not strictly necessary.

1500 SPM would require about 100 wagons loading a full set every minute, depending on your bonus. Your factory would need to be large enough to handle the incoming trains from this mining op every minute. To make it arbitrarily scaleable, you could make each processing step vertically expandable. A step at the very end could be used to collect any unused ore / intermediates.

Basically a processing step would be like furnaces or assemblers pulling from the wagon and putting it back into the wagon.

Once you have the base factory down, you can just copy paste them vertically to expand and go out looking for more patches.

I believe this would be very UPS efficient, but not sure. Likely there would be some idle assemblers / furnaces in this model compared to one dealing with a compressed belts which might be a UPS issue. I don't really know, tbh.


But, yeah. With belt weaving, you can get 2 extra sides or so. With a train, you got 40 sides! Really simplifies things. Plus you've got random access belt sides! No side switching required.

I posted this in the factorio speedrun forum as well for 100%. I suspect it could seriously shorten the time there as blueprints and megapatches are allowed (and practically no biters).

https://www.speedrun.com/factorio/thread/udz80/1#b1hov

Here's a basic idea for the factory:

https://imgur.com/a/TtoupDD

I've just done the rs/gs to start off with here. Each horizontal line will be a complete set of processing stops: plates, gc, gears, belts, inserters, red science, green science, steel, pipes, rc, engines, sulfur, blue science, sticks, rails, modules, bricks, furnaces, violet science, bc, red engines, batteries, frames, lds, yellow science. rcu. jetfuel is all fluid so mabye not train fed. white science you'll want to pull and deliver to the labs like any other science. Again, no funky belts required as you have 40 different spots to put stuff.

To expand for more production, you just copy/paste a full line. You'll want to make sure you have a queue of trains waiting to get into the factory before you do this though, so you are balancing adding production with adding factory capability.

Note that you'll have trains queued at every stop and in every stage of proceeding to full science.

This approach is somewhat orthogonal to the long trains discussion. I created a new thread here - viewtopic.php?f=194&t=100193

[Redpossum]

Is there a problem with loading and unloading trains from both sides at once?

[mrvn]

Is there a problem with loading and unloading trains from both sides at once?

That's what many people do.

Even with fluid wagons you can do it, or basically must if you want 3 pumps per wagon. Can't place 3 tanks per wagon on one side only for more than one fluid wagon.