Afferent/Efferent Rail Design

Smart setups of railway stations, intelligent routing, solutions to complex train-routing problems.
Please provide - only if it makes sense of course - a blueprint of your creation.
Dry Hairy Tree
Filter Inserter
Filter Inserter
Posts: 319
Joined: Wed Jun 08, 2016 10:51 pm
Contact:

Afferent/Efferent Rail Design

Post by Dry Hairy Tree »

As biological systems often prompt better human design: I've been considering the way our central nervous system relays information and how that might relate to Factorio game play.

Very roughly... Nerve signals are relayed via incoming (afferent) and outgoing (efferent) neurons. These do not intersect. Their respective signals are transmitted (across synapses) to each other via electrical impulses. This is part of the design for a system that is at once automated (parasympathetic) and responsive to it's environment (sympathetic), while dealing with enormous data.

Sounds good right?

For Factorio - trains become afferent/efferent nerves, and logistic systems are the signals across them. Trains thus do not intersect as you see in my first attempt at the concept below.
AfEf 1.jpg
AfEf 1.jpg (78.91 KiB) Viewed 12092 times
As the tracks do not intersect other tracks there is potential for back to back trains on such lines acting as a 'belt of wagons' for very high throughput. Stackers are not required as the entire track is potential stacking space.

Afferent and efferent, incoming and outgoing. The above example is a circuit factory. There's two trains in and one out, although (100M) copper is onsite for now so, eventually there will be three trains in, one out.

Externally (away from the factory central) the handling of multiple outposts can be done in a manner inspired by another biological phenomenon - fungal hyphae.

Hyphae extend and detect resources, and then proliferate in the area till that specific resource is gone, then they're gone from the area, and another species moves in - for another resource e.g. brown rot fungi take the cellulose then white rot fungi move in for the lignin.

The trains likewise do their thing separately. While these networks will branch out - mine copper on one side of the base and iron on the other e.g west copper vs east iron vs north uranium vs south coal.... then a messy entanglement of trains does not occur. Deplete one specific resource then move in for the next after the previous has gone. Or, to use another biological term, it is ' functionally extinct'.

Dry Hairy Tree
Filter Inserter
Filter Inserter
Posts: 319
Joined: Wed Jun 08, 2016 10:51 pm
Contact:

Re: Afferent/Efferent Rail Design

Post by Dry Hairy Tree »

A quick schematic reveals that, on paper at least, the concept is sound.
AfEf 3.jpg
AfEf 3.jpg (56.59 KiB) Viewed 12091 times

Dry Hairy Tree
Filter Inserter
Filter Inserter
Posts: 319
Joined: Wed Jun 08, 2016 10:51 pm
Contact:

Re: Afferent/Efferent Rail Design

Post by Dry Hairy Tree »

I got the big brother mod and the helicopter mod and took a bunch of solar powered radar for a very long flight to reveal an area of map that takes 50 solid fuel for the helicopter to fly across.
AfEf 4.jpg
AfEf 4.jpg (164.54 KiB) Viewed 12086 times
The map exploration reveals a typical spread of resources that accommodates the proposed separation of mining enterprises to polar opposites (and then circling so as to cover the territory while avoiding each other). As you can see there is variation for certain areas but nothing particularly problematic.

It also shows how, when planning to go large, it pays to take a good look around. I moved from my initial wee base and it's pathetic ore surrounds (top centre); to '35 solid fuel SSW' (bottom left of map) where I'd discovered what I wanted - my first triple digit ore patch - it was exactly 100M copper. I've since discovered an iron one over 200M (which is unfortunately 50 solid fuel away. No problem. I have well over half a billion ore each of copper and iron within a short distance from my base. I imagine it gets even richer the further out and will take a look at some stage (Rail settings, RSO mod).

Why would I go back home and hook up the 3M's, 8's, 16's...

The initial base is retrofit and makes boxes of robots and nuclear facilities till it runs down. I'm here in my new home, surrounded by mega-resources, with afferent/efferent neurons firing over my afferent/efferent designs.

vanatteveldt
Filter Inserter
Filter Inserter
Posts: 945
Joined: Wed Nov 25, 2015 11:44 am
Contact:

Re: Afferent/Efferent Rail Design

Post by vanatteveldt »

Cool idea. It is somehow similar to a hub design with dedicated lines. I like the idea of guaranteeing that each track has a unique commodity, makes a number of things a lot easier.

I was weaving it into an existing game but I think I'll start over without biters so I'm not space constrained.

See viewtopic.php?f=5&t=51685 for some discussion and station dressing for the hubs/synapses

Dry Hairy Tree
Filter Inserter
Filter Inserter
Posts: 319
Joined: Wed Jun 08, 2016 10:51 pm
Contact:

Re: Afferent/Efferent Rail Design

Post by Dry Hairy Tree »

I like your ideas/designs. The discussion is interesting too.

Unfortunately bottlenecks will always occur with multi-train choke points whether you selectively mine different sectors or not. It's getting lots of trains in and out without interference. I've had a play around for a multiple line input you can see the bones of below.
AfEf 5.jpg
AfEf 5.jpg (223.65 KiB) Viewed 12028 times
It's easily powered and covered by roboports. I'll use these or something similar to run sideways through the center of smelting areas placed as in the schematic I made. By running in through the centre sideways: I wont interfere with other lines in the plan, and it will alleviate the issue of the space this thing takes. The 'clover' can be tightened up 4 squares but it doesn't look as good. It can be adapted to any number of lines but roboport cover limits things.

I started making this base before I made the schematic. As always with retrospect there's room for improvement. The schematic's all good things go where they should, the base, well, I got close... :D

Dry Hairy Tree
Filter Inserter
Filter Inserter
Posts: 319
Joined: Wed Jun 08, 2016 10:51 pm
Contact:

Re: Afferent/Efferent Rail Design

Post by Dry Hairy Tree »

Of course single sided unloading is not conducive to throughput, it'll definitely need a rehaul.

vanatteveldt
Filter Inserter
Filter Inserter
Posts: 945
Joined: Wed Nov 25, 2015 11:44 am
Contact:

Re: Afferent/Efferent Rail Design

Post by vanatteveldt »

So if you have 10 iron outposts, would you make 10 single-train unloading bays to prevent crossing tracks? I feels like at some point it would be better to merge some of them, and it seems that a merged track can still provide enough throughput to keep 6 stations busy...

Dry Hairy Tree
Filter Inserter
Filter Inserter
Posts: 319
Joined: Wed Jun 08, 2016 10:51 pm
Contact:

Re: Afferent/Efferent Rail Design

Post by Dry Hairy Tree »

Ok. The external (ore) lines vary to those internal in the factory. In the factory is a neuronal model. For the ore it's a (fungal) hyphal model. Ore lines head straight for, then branch out to, resources. As many outposts as it takes to have one station (of size you determine) running continuously should be the max side branching of one ore line. It get's made so trains never wait on each other unless in the specific queue they should be in. With fungal hyphae they branch consistent with the volume the main thread can transport. There's definitely a pattern of more branching where more resources are found, however, we're limited in that these biological systems work in three dimensions and we've only two (sans underground pipes/belts).

I've been rethinking and designing many train stations for ore. I'm thinking maybe the ore trains would be best as double headers using a drop-off station designed similar to below.
Af Ef 6.jpg
Af Ef 6.jpg (230.76 KiB) Viewed 11985 times
A dedicated line in with trains stacked and waiting, a dedicated line out. Each line never intersecting. Then add blocks of smelting, then another (or more) station/s similar to this. More smelting...

I'm interested in determining how many module & beaconed smelters a certain sized station can serve - with all inserter upgrades. Any idea?

Dry Hairy Tree
Filter Inserter
Filter Inserter
Posts: 319
Joined: Wed Jun 08, 2016 10:51 pm
Contact:

Re: Afferent/Efferent Rail Design

Post by Dry Hairy Tree »

Been testing station turnaround times. Loops beat double headers hands down. The trick is to stick a signal between every wagon so one train enters at the same time another is leaving as pictured below.
AfEf 7.jpg
AfEf 7.jpg (194.16 KiB) Viewed 11974 times
This delivers 8 wagon trains (16K ore) roughly every 14 seconds. With the double header trains I couldn't get it below 20 seconds.

I'll have 300 smelters per side for a total of 600 serving the station. 2 1/2 double rows with beacons, roboports, 2 1/2 double rows with beacons, more roboports - per side. This gets excellent roboport coverage which I'm sure will be crucial.

Each smelter receives 26.66 ore to process which at a rate of 1.667 per second takes approximately 16 seconds to process 16K ore and subsequently produce 19.2K plates or 72 K per minute. I'm not sure what kind of numbers you need but that's some (theoretically) solid output if you start placing multiple stations. Steel will work with the same 600 smelters only use a 4 wagon train.

At each step of the way I've rounded in favor of more supply.

Throughput should not be an issue if I set it up right. That'll make a welcome change. I made one bus - one! Set up a huge pretty factory with multiple blue belts then it came to making circuits - rubbish :D

Dry Hairy Tree
Filter Inserter
Filter Inserter
Posts: 319
Joined: Wed Jun 08, 2016 10:51 pm
Contact:

Re: Afferent/Efferent Rail Design

Post by Dry Hairy Tree »

AfEf 8.jpg
AfEf 8.jpg (517.21 KiB) Viewed 11960 times
Smelting stations like this. Only one beacon short, and some roboports are missing from the top - so it doesn't connect to the logistics network yet.

vanatteveldt
Filter Inserter
Filter Inserter
Posts: 945
Joined: Wed Nov 25, 2015 11:44 am
Contact:

Re: Afferent/Efferent Rail Design

Post by vanatteveldt »

Cool ideas. Answer to just limit plants to one station capacity.

I'll combine with my hub idea and see how far I get :)

Do you build with creative mode of do you do it the hard way?

Edit: did you consider 1-4-4-4 trains? :) viewtopic.php?f=194&t=51751

Dry Hairy Tree
Filter Inserter
Filter Inserter
Posts: 319
Joined: Wed Jun 08, 2016 10:51 pm
Contact:

Re: Afferent/Efferent Rail Design

Post by Dry Hairy Tree »

Dang I replied already, but obviously did not hit reply.

Those trains do not unload any faster, just go in a circle. With the large number of smelters needed there's no need to shorten station lengths I'd have to have 20 rows per side at half the length... That said they're very cool out of the box thinking which I've always appreciated.

Built the station the hard way. Speaking of hard way - I've started again....

The huge map reveal of lands I was not using was causing serious lag when autosaves happened and typically felt sluggish compared to normal. Also, the plans came together after half the base was already built and I can't be bothered tearing it all up.

So I got Arumbas fast start, jumped in my car, and headed north immediately - seeking three digit ore patches. Took about an hour... (trees) but found triple digit copper, iron and coal in the same area by some oil so set up power and am about to start radar scanning for stone/uranium I hope is reasonably close. This is much better than schlepping goods across the map with the helicopter, just move immediately and start the base where you plan to stay. Aliens on peaceful cos I just couldn't be bothered with them as I'm testing new ideas. I find dealing with biters ridiculously easy but they cost too much in time.

Today I'll put in a building store - a small factory making the things required for making a big factory. And I'll do some more math...

Dry Hairy Tree
Filter Inserter
Filter Inserter
Posts: 319
Joined: Wed Jun 08, 2016 10:51 pm
Contact:

Re: Afferent/Efferent Rail Design

Post by Dry Hairy Tree »

AfEf 9.jpg
AfEf 9.jpg (48.88 KiB) Viewed 11831 times
Progress. The hardest part is removing trees. I had to bot-mine rocks for smelters and some rail at first. Got over 2000 and was quite proud of myself then finally found a stone patch of 40M or something ridiculous. Since dumping that stone I mined I have another 186 landfill. That's not a lot of landfill but it is a lot of rocks.

I'll get this first half running - put in the mining, a refinery and some preliminary power. Then I'll use the facility to create it's own modules/beacons for itself, and get some labs in to start picking up later-game robot/train/mining upgrades. Fair bit of work to go but it's all good really.

After the schematic is done (note this varies to schematic one, and variants are certainly possible), the next figuring in the 'flow of things' is what goes on what train line. I numbered the trains involved and wrote lists for deliveries. e.g. I will make many of the components for science pack assembly at the labs themselves so... I'll ship in plate, steel, circuits, modules, bricks, coal and refinery products. Everything else is made from these. Two lines required, one from the refinery, one from the circuit factory.

Shame life keeps interfering with my Factorio...

But there's progress. :P

Dry Hairy Tree
Filter Inserter
Filter Inserter
Posts: 319
Joined: Wed Jun 08, 2016 10:51 pm
Contact:

Re: Afferent/Efferent Rail Design

Post by Dry Hairy Tree »

Here's an outpost. There's 850 miners to service 94 million iron. Many roboports to go in yet. And, can't turn it on yet, too much power!
AfEf 10.jpg
AfEf 10.jpg (65.12 KiB) Viewed 11811 times
I think this might be enough to serve one station alone depending on train travel times (it's very close), but I have only set it up for four trains, via two stations, so we shall see.

See the branching occurs at the resource (fungal tactic). Entering at the right the first station is on the right hand track, and has room to queue a second train. The track on the left goes to the second station, and a merge point only occurs after this where, if a train is held up, it is because another train has beat it to do its own task, in other words, not a problem.

Watched another video of an expert unclogging his train jams last night... :mrgreen:

Power is always the pain (for me) going from small to bigger. Steam is demanding to feed and a bit weak, solar is boring, but that kovarex process is not cheap! I make do with silly amounts of steam till the (proper) labs go in. Still, even if it was cheap, got to get 40 x 235 or it's pointless having it.

At the theoretical max input of 72K per minute ore to the station, it would take over 21 hours to deplete this patch.

mrvn
Smart Inserter
Smart Inserter
Posts: 5684
Joined: Mon Sep 05, 2016 9:10 am
Contact:

Re: Afferent/Efferent Rail Design

Post by mrvn »

I wonder how far you can take this without having to cross tracks. Have you considered everything you need to build and launch a rocket?

Although after the first few production steps the volume goes so far down you probably won't need dedicated tracks anymore. So the solution might be to go 3 ways: central mixed ring, cross free lanes for production and branching for mines.

vanatteveldt
Filter Inserter
Filter Inserter
Posts: 945
Joined: Wed Nov 25, 2015 11:44 am
Contact:

Re: Afferent/Efferent Rail Design

Post by vanatteveldt »

Not to threadjack, but this is what I think is needed for all science minus rocket science:

Of course it depends on exactly what you want to produce separately, I guess making the LCU and RDS in situ also makes sense given the relatively low volumes, and you could also make all the science packs in different plants. But with the assumptions that you have separate locations for mining, smelting, circuits (x3), LCU, RDS, rockets, and science, (and that you use trains for refinery products, ie rocket fuel, plastic, and sulfur or sulfur gas) you get something like this:
dependencies
So you can't avoid some crossing (unless you want to duplicate plants, but that sort of defeats the purpose of creating separate plants anyway...), but it's not too bad IMHO; and in any case straight crossings are better than full intersections.

Maybe you could create a "crossing" by cutting one of the lines, placing stations and loops at either end, and have bots transport the good over the other track? That would increase latency (and cost), but should not affect throughput of the crossing line (so you should be able to get maximum throughput out of both lines):
non-level crossing
(OTOH, a single station cannot unload a saturated track anyway, so with some level crossings the track throughput still won't be the bottleneck?)

Another way to avoid that might be to run tracks through the 'synaptic gap' of the plants, ie the copper to science track could go through the green plant (effectively, you replace a rail-rail crossing by a rail-bot crossing), btu that doesn't seem very scalable since the throughput of the copper -> science link is now limited by how much space you leave in the plant. Alternatively,

(I would also think about a passenger / supply train, presumably on separate track, but again it's not such a problem if that crosses the production tracks somewhere.
dot code

(@DHT: I can move this to a separate thread if you wish, but it seems relevant?)

mrvn
Smart Inserter
Smart Inserter
Posts: 5684
Joined: Mon Sep 05, 2016 9:10 am
Contact:

Re: Afferent/Efferent Rail Design

Post by mrvn »

Science must come together at the lab so it changes nothing if you produce it separately (graph wise).

But steal can be produced from iron instead of from mines. Maybe that helps.

Green circuit boards I often produce in place. It's such an easy setup and you need all the output right there for red/blue circuit boards. Not sure if that helps though. Maybe for science since they already need Iron and Copper.

As for your idea of a "crossing" by cutting one of the lines. I would try two other things first:

1) Pass goods through a station. For example red circuit boards are needed for blue circuit boards and science. Why not have the red trains stop at blue and then go on to science? You can do that with green too. A whole lot of crossings just went *puff*.

2) Instead of running through a gap in a factory run a track between the factory and the loading/unloading station. Unload onto underground belts and pass them below the other track(s). This can be extended quite easily by moving the unloading station and adding more underground belts. Not perfect but better than moving the factory.

I also love BOBs configurable inserters. You could have 42 inserters total unloading per wagon. But I find 6 on a side usually does the trick if you set them up to minimize head movement. Might even be the fastest. That gains you about 9 times the speed over vanilla setup and fills a belt to ~70-90% depending on your stack bonus, certainly more than 50%. So more than 2 unloading stations makes little sense.

Dry Hairy Tree
Filter Inserter
Filter Inserter
Posts: 319
Joined: Wed Jun 08, 2016 10:51 pm
Contact:

Re: Afferent/Efferent Rail Design

Post by Dry Hairy Tree »

You are most welcome to post ideas here this is a work under construction and the feedback forces me to think harder about what I'm doing.

The station crossover/synaptic cleft you try to describe/design is in the design already. If you look you'll see much of the crossover of your diagram has been eliminated (well, all, there's no crossing over). You can actually mine dual resources from one area and never cross over, given the entire map - there's no excuse for it. I designed something almost identical thinking the same stuff. Robots bridge synapses, always, BUT, if traffic is very low, and belts can take it, why not? That darn rail spacing grrr, i'd have chest to chest stack insertion otherwise.

Everything flows to two central points, labs and building/military. Greens go via bot to red, then both to blue. Handily where all get picked up for transport. Sure there's still plate/plastic bars etc getting moved but the majority of resources are greatly reduced when they get turned to steel/circuits and modules. There's no trains moving circuits to circuits, bots do it.

Rocket production is simply not an issue. There's three trains leaving the refinery, the one direct to the lab will carry solid fuel/plastic as well as some lube/batteries. Lube and batteries = 2 wagon maximum. The train from circuits/modules carries steel and plates, speed modules, blue circuits.... Job done.

The japanese use slime mold for transport design. They make model landscapes and place resources where cities go and let the little suckers go to work. Taking elevation, tunnelling and distance all into account the slime mold finds the fastest and easiest route to take and make.

I learn from the slime mold, the fungus and brackets, the clattering crawly things, mildews and maggots. :D

vanatteveldt
Filter Inserter
Filter Inserter
Posts: 945
Joined: Wed Nov 25, 2015 11:44 am
Contact:

Re: Afferent/Efferent Rail Design

Post by vanatteveldt »

I suppose the big question is where you want to place the challenge :). You could also move everything by bot, but then it would be a very boring thread (and base).

I agree that you can eliminate all crossings by designing it as a tree structure with the labs as root node, but for my base I think I will force all identical resources to be distributed by a single hub, thus creating some crossings. I've started building but so far I've only built my bootstrap base to do necessary research and build what I need to build the tracks, smelters, circuit plants and final building store.

{Note that the design above is not for a normal station, but simply to create a non-level crossing of two tracks. In a way it resembles the portages that were used to connect water without a canal, e.g. by carrying cargo overland at Suez. }

Dry Hairy Tree
Filter Inserter
Filter Inserter
Posts: 319
Joined: Wed Jun 08, 2016 10:51 pm
Contact:

Re: Afferent/Efferent Rail Design

Post by Dry Hairy Tree »

Well, power was certainly an issue and I've already devoted a rant to it... :D

I put in 160 more steam engines to total 200 and turned on the outpost and one station and within a minute had a rolling blackout - no coal. The darn things were being fed red belts. So, I had to face facts - my pre-0.15 building store just doesn't cut it any more. I need production and high tech dribbling in from early on so I don't spend a day retrofitting a base for higher level tech and a nuclear plant...

Basically, I went too big too fast and power didn't match the scope/tech. I was impressed by how fast I was going big but... lol. Now I have nukes, kovarex, and a crazy new kovarex design. I'll tidy it up and show y'all later. It was getting late...

I'm gonna level that starter base with nukes when I'm finished with it. And next game design a new 'fast start' base that includes 0.15's shenanigans.

I'm rethinking the central hub of it all. the Mickey Mouse ears are kinda cool, but they take a lot of space. I'm trying to avoid turning the lines at the ends of the smelting area as the configuration looks like a penis. BAAA HAAA. Still funny after all these years. Running straight through seems best fit at first, but with tanks being incorporated, more thinking is required. Still, best fit is only theoretical, limited by our imagination.

Preliminary testing of the iron outpost/station with nowhere near enough bots/ports shows it running so freaking sweet. I put them on a timer as if they had all the inserter upgrades to see how it looks. The flow of trains is a thing of sheer beauty - but one outpost probably won't cut it even with a lot more bots. I started with the outpost stations full and it kept up for a while... it was so close I'm confident two outposts, four stations and eight trains would almost certainly be enough for one smelting station.

You mentioned roads earlier - get a helicopter! So superior for a big base, my only real complaint is it needs floodlights.

I'm interested in seeing how your plans turn out. I see that some crossings might help me in some places, and some multi-station type deals on one line... but I've stuck to my no intersection rule as it is the key to uninterrupted throughput (that and supply).

So many design ideas the build is going slower as I explore options hoping to ultimately build a better prototype.

Post Reply

Return to “Railway Setups”