Explain intersection based vs. loop based rail networks

[hitzu]

This is a nice problem. I found an easy solution: double headed trains needs to be always an odd number of wagons. 1, 3, 5 wagons (more don't make sense). In that way, you can keep them symmetrical!

I don't get why an even number of wagons makes them less symmetrical.

[SpeedDaemon]

Ahh, yes, I have made this kind of experience lately. This is a nice problem. I found an easy solution: double headed trains needs to be always an odd number of wagons. 1, 3, 5 wagons (more don't make sense). In that way, you can keep them symmetrical!

While single headed trains (one direction only) make more sense with even numbers of wagons: 2, 4, 6 (more makes also no sense).

This was actually a solution I considered (and it will work for any number of wagons; if you have two, for example, they just have to have the same setup). That means, though that all of your trains and stations essentially have to be 2x the size, and the redundant loading areas mean you're storing potentially thousands of expensive items in loading buffers that you don't really need (and there were some other more complicated problems).

[Gus_Smedstad]

I just converted my copper train to a double-headed setup. It did free up some valuable space, but I was surprised at how much slower the train was. When testing a train schedule I'll ride all the way through it, usually forcing it to move to the next station once I've verified it lines up where it should. I'll often hover over the train as it's driving, watching the speed - and the numbers are much, much lower for a 2-headed train with 2 cargo cars than a single locomotive with 1 cargo car. Obviously the extra locomotive adds much more weight than I thought.

I know in real life, trains often use multiple diesels for extra power, but I have no idea whether full power is available in reverse. Logically it should be, since the actual drivers are electric and don't care about direction the way a combustion or steam engine does, but I don't know for certain that it is.

[MeduSalem]

MeduSalem is incorrect. It turns out that Y-intersections are not enough for many situations. At least in part because the oil train generally makes the rounds to many different wells, picking up a few barrels from each, rather than shuttling between a mine and a smelting station the way the iron and copper trains do. Once depleted, oils wells produce a trickle, but that trickle never completely vanishes the way ore mines do, so it's worth keeping the stations. So while ore mines can get away with Y-intersections, oil depots and branches of the main line need full T intersections. Else trains up end traveling through odd stations just to turn around.

I never use trains to transport oil at all. While I like fiddling with almost everything else in factorio, I highly dislike fiddling with transporting oil in barrels. I just hate that, so I always use pipes and small pumps to get the oil to the refinieres even if it means laying pipes through a huge wasteland (with personal roboport+blueprint this isn't an issue anymore). Guess that's why I'm not stumbling across any problems with having Y-intersections only as I don't need any trains going from one outpost to another to pick up some more stuff.

Also I solely rely on solid fuel for my power production (I hate solar energy for the huge space it wastes all while looking boring and being a shallow feature due to how there's no thoughtful maintanence mechanism involved. Plop&forget = boring). So I don't want to rely on trains and barrels for delivering the oil I need, because if there's a power shortage this might lead to a fatal chain reaction where I don't have enough power to create more barrels which would be needed to transport more oil. Not going to take the risk. Also I don't like the barrel concept in the first place, I'm still waiting for the official implementation of tank wagons which can be filled with any fluid desired. Maybe then I'll give it a shot. (Yeah I know there's a mod to do that already, but in generally I avoid using mods in alpha games because they get broken constantly and often the modders lose interest in keeping the mods updated due to the ongoing frustration of having to permantely keep stuff up-to-date.)


But back on topic, I perfectly understand that my method of using Y-forks only obviously doesn't work in every situation. If you have like a dozen oil outposts and the train stops at each single one then they need be able to go there of course, then you won't get around using at least T-junctions. In huge factories where almost everything is transported only by rail exclusively the same thing applies.

[Gus_Smedstad]

While I like fiddling with almost everything else in factorio, I highly dislike fiddling with transporting oil in barrels.

I enjoy how fiddly it is. It's the only transport that requires anything beyond "load here, unload at base." Managing the empties and having to fill the barrels adds a little something. I'm aware it's probably not as effective as just using pipes and maybe the occasional pump, but I build them to have another reason to build a rail network.

Solar energy I like because it's an investment, and I make a point of setting up off switches for my steam engines so I'm burning as little fuel as possible. Coal's needed for plastics and sulfur and isn't unlimited the way oil and solar are, and oil's always in short supply because gas is needed for plastics and sulfur again. Which are key to the advanced electronics, which always end up being the bottleneck for the highest tech stuff.

Power still requires periodic expansion even if it's mainly solar. This ends up being "maintenance" of a sort, more so than building additional steam engines, which is easier because the energy density is much higher.

I looked at the oil tanker mod, and it looked like it would probably be erratic. Particularly the bit about "connection points moving around, maybe one will work."

I wish there were more reasons to transport things by rail, but for the most part it feels like it's only really valid for raw materials, and that because they're scattered and they deplete. Finished goods generally travel short distances and make more sense to move by belt or logistic robot.

[MeduSalem]

I wish there were more reasons to transport things by rail, but for the most part it feels like it's only really valid for raw materials, and that because they're scattered and they deplete. Finished goods generally travel short distances and make more sense to move by belt or logistic robot.

I mean yeah, it's not on topic, but I'd also wish for that.

Probably they are never going to implement it but I'd actually like to see settlements that are found very, very rarely on the map, far, far away from your starting point you could start trading with.

I mean it wouldn't fit into the currently very distinctive theme of being "stranded alone on an alien planet" anymore (which would still be valid for the first part of the game, but not anymore once you made contact), but it would actually give a reason to further mine resources, manufacture goods in your factory, deliver them by train into those growing settlements (which could be other humans or aliens who try to colonize the planet) and they may give you something in return you may need yourself.

I don't know if someone ever made that suggestion, but that's one of the things I would have liked to see for the endgame, next to the space stuff, which is hopefully going to be way above and beyond everything else.

Settlements/Colonies would at least give it a major reason to keep on exploring, expanding and producing goods and to use the train network more extensively. Currently there's like no reason to expand because beyond the dull and boring wasteland filled with nasty bugs is just more of the same dull landscape and there's nothing to waste the resources on except some self-set goals which is an excuse I can't get myself acquainted with after some weeks of intensively playing a new update after which I have to take a break eventually to free my mind and catch up on other stuff.

Currently there's not really a point in getting all thrilled about the rail discussion when basically most people finish their games before they even depleted the resources in their starting area, never actually needing to build a rail to get somewhere.

[Zhab]

While I like fiddling with almost everything else in factorio, I highly dislike fiddling with transporting oil in barrels.

I enjoy how fiddly it is. It's the only transport that requires anything beyond "load here, unload at base." Managing the empties and having to fill the barrels adds a little something. I'm aware it's probably not as effective as just using pipes and maybe the occasional pump, but I build them to have another reason to build a rail network.

Solar energy I like because it's an investment, and I make a point of setting up off switches for my steam engines so I'm burning as little fuel as possible. Coal's needed for plastics and sulfur and isn't unlimited the way oil and solar are, and oil's always in short supply because gas is needed for plastics and sulfur again. Which are key to the advanced electronics, which always end up being the bottleneck for the highest tech stuff.

Power still requires periodic expansion even if it's mainly solar. This ends up being "maintenance" of a sort, more so than building additional steam engines, which is easier because the energy density is much higher.

I looked at the oil tanker mod, and it looked like it would probably be erratic. Particularly the bit about "connection points moving around, maybe one will work."

I wish there were more reasons to transport things by rail, but for the most part it feels like it's only really valid for raw materials, and that because they're scattered and they deplete. Finished goods generally travel short distances and make more sense to move by belt or logistic robot.

I agree with everything you said. To me it is steam engines that is the cheap, easy and boring option. The only maintenance involve is the occasional need to find a new coal patch. Which you usually stumble upon by accident while looking for copper/iron/oil.

A good argument to use steam thou is that it does significantly contribute to pissing off the aliens more in early to mid game. Let's face it, aliens are not a problem in end game. At the very least not from a base defense point of view.

And I'm definitively all in for extra reasons to explore the map by train, car, tank or otherwise.

[OBAMA MCLAMA]

Thats a lot to read. I didnt read everything but I will throw a couple words out I suppose.

With a double headed train... nooooo neeed for odd. The only train that you need to worry about going in the right way would be your engineering train and oil. In which case you just use some filters and set it up so it doesnt matter which way the train goes in. Its easy.

And.... steam engines being the easy way out huh...... they consume far more resources than a solar panel. I run bases 100% steam powered. If your steam engines fail to receive coal, it could take hours to fix it depending how much planning you have done.
And its not just go find a new coal patch, compared to the solar panels being oh... need more power... stamp stamp done.
Could say more but im on my phone.

[Gus_Smedstad]

So, this happened.



This is an example of the issues with intersections, which I believe you can avoid with rotaries. Namely, unexpected deadlocks that arise because you need more signals than is obvious. If we number these trains left to right as 1-3, trains 2 and 3 are just passing each other, neither is actually in conflict. But train 2 is in a block that includes rail heading northwest, rail that crosses the opposite rail, and that locks the block that train 3 wants to enter. Train 3 is in a block that includes track that crosses the other rail, a block that train 2 wants to enter.

The issue is that both are on blocks that include switches and a track crossing. The switches have to be in separate blocks from the track crossings, or you get deadlocks like this.

If the tracks are at all close, it may be impossible to separate switches from crossings. If you look at the switch NW of train 3, it's impossible to put a signal on the west-bound track before the crossing, because the end of the curved track piece is part of that crossing.

[MeduSalem]

So, this happened.

picture

Nice... xD

That is another perfect example why no space between tracks (at least on the diagonals in this case) is really, really bad, because then you are actually forced to have full train-length blocks in between two intersections so they can completely leave each intersection before entering the next and even then you might end up with deadlocks when multiple trains get there at the same time blocking each other's exits.

I don't even know if a rotary could save that because it would be pretty tough to make one in the space given and that allows for all 4 exits. So you'll have to make one huge intersection out of that with chainsignals everywhere except the the 4 exits, which turns that thing into a huge bottleneck due to not being able to place the neccessary signals. It's a mess. I'd tear that down immediately. xD

[Gus_Smedstad]

I don't even know if a rotary could save that because it would be pretty tough to make one in the space given and that allows for all 4 exits.

It's slightly larger than the 3-way intersection, but it fits, and it wasn't a strain. The other track remains a Y-intersection.

I believe the rotary cannot get a deadlock, because each switch is in its own block (a couple of signals are hidden by the train here).

[SpeedDaemon]

So, this happened.

Big Pic

There are actually two problems there. The lack of some signals makes it worse, but those two intersections are also too close together. You need at least enough space between the exit signal of one intersection and the entrance to the next for your longest trains, or even a "non-deadlockable" intersection can deadlock. You could probably solve that problem now by chaining signals through both intersections, at the cost of having some trains stop when they wouldn't really need to.

[hitzu]

So, this happened.

You forgot at least 7 signals. No wonder that it happened.

[Gus_Smedstad]

You forgot at least 7 signals. No wonder that it happened.

It turned out that the two spots near train 2 you highlighted are not legal locations, because they touch other rails. Putting signals there resulted in flashing errors. I tried fixing the deadlock by placing signals there, and failed. Which is why I tore it up and put in a rotatary.

This relates to what I said earlier - if the rails are too close, safe signalling is impossible, and the number of additional internal signals you need is often unintuitive.

[hitzu]

This relates to what I said earlier - if the rails are too close, safe signalling is impossible, and the number of additional internal signals you need is often unintuitive.

Therefore making at least a 4 tiles gap is essential. Just place a signal before and after every time when two rails connect each other. The last signal should be a normal signal, but only if there is enough space behind it for every train.

[OBAMA MCLAMA]

Yea.. your rails are way to close to each other, Need to be able to fit a roboport in between the rails. (space wise)
the deadlock you see, could have been avoidable. It wasn't a situation where it is always going to deadlock.

[Gus_Smedstad]

Which points to an advantage of rotaries I hadn't expected, that rails can be arbitrarily close together and still work, whereas Y / T / Plus intersections require a minimum separation of track. Which is unimportant in the outskirts, but eats up space near the primary base, since the space between the tracks is difficult to use effectively. I suspect that the space used by the separation is more of an issue overall than the slight increase of size of a rotary over an intersection.

[ratchetfreak]

the blue circled signals should be chains to avoid the trains waiting where they do. (Following my rule of "at every signal where you don't want a train to wait at go back and change the incoming signals to chains")

The red points is where your problem is (though no the only ones). Their absence makes the actual deadlock because the trains are waiting on each other putting the signals there will split up the block so the trains won't actually interfere with each other.

[Gus_Smedstad]

I didn't quite grasp "all points where tracks cross in any way must be in individual blocks" when I started laying track. I just put signals at the start and end of the intersection, and figured that it might not be efficient, but it wouldn't deadlock if it treated the entire T-intersection as a single block. I was incorrect about that, and I started adding internal signals as I encountered problems.

As I said before, there are points which require signals which cannot take them in that layout, because the tracks are too close. it's the curved pieces that are the root problem, since they're long, and one piece can easily cross tracks in two places if the tracks are at all close, which makes safe signalling impossible.

It's interesting how deadlocks can happen even if train traffic is very light, short trains and not many of them with lots of track. Sooner or later trains will go through the same corner at the same time.

[hitzu]

For that layout you can enlarge the whole intersection, making some internal blocks more than one train lenght and basically turning the T-intersectiom into three Y-intersections. That is potentially can reduce chances of the deadlock.