## 3 and 4 way intersections

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.
Avona
Long Handed Inserter
Posts: 62
Joined: Sun Dec 26, 2021 8:50 am
Contact:

### Re: 3 and 4 way intersections

Hello!
PavelH wrote:
Sun Oct 01, 2023 4:37 pm
1. Is there some intro writeup to intersection design? I see a bunch of patterns repeated in multiple of the intersections here, some of which make sense in retrospect but some of them are a mystery to me.
I have not seen any advanced intersection writeup. Most tutorials go over the signals. I'd suggest asking questions in the Factorio discord server in the #train-help channel. There are many experienced people there who can answer many of your questions. That being said, I will go over the rest of your post.
PavelH wrote:
Sun Oct 01, 2023 4:37 pm
1a. A particular pattern that I don't understand is parallel lines on the non-intersecting turns - since both the input and output are single lane, I don't understand how this helps throughput. Is it somehow related to how the game selects which train has priority to enter the final block?
Parallel lanes on non-crossing turns are for consistency of throughput. They buffer trains against the randomness of the input. They are not as effective as buffering before and after crossings, of course.
PavelH wrote:
Sun Oct 01, 2023 4:37 pm
1b. Another pattern I don't understand are long non-signaled section that are not immediately after a crossing, or are much later than needed for the crossing. What role do these play? I understand having too dense signals causes throughout to drop due to trains not being able to speed up, but this seems excessive.
A specific example of this would be "parallel multicross turbo", where the blocks preceding the exit are at least 12 carriages long, with some other blocks also being seemingly unnecessarily long
There are two things that could be going on here. One is that the exit blocks after crossings don't actually have to be directly after the crossing. They can be offset from the crossing at intervals of full train lengths if they are consistently a certain length.
In the Parallel Multicross Turbo, and a few of the others, they utilize high throughput merges. Regular merges get about 27 tpm for 2-4 trains. High throughput merges weave together traffic so that trains will merge at high speed instead of coming to a stop before merging. They can get about 33-34 tpm with this sort of merge.
You also said "having too dense signals causes throughout to drop". In general, the more dense the signals, the higher the throughput. After a crossing, the exit blocks must be the length of your longest trains to prevent deadlock.
PavelH wrote:
Sun Oct 01, 2023 4:37 pm
2. Is there a reason all intersections are using signals in one line? Personally, I have an offset of 1 tile between the "in" and "out" signals, because that enables the straight section to be a rotationally symmetric blueprint, and (according to the indicator when placing the signals) the train stops in the same place. Does this matter at all?
Signals are used in one line for mirror symmetry I assume. I also offset 1 tile for my straight sections. It doesn't really matter. The blocks should be sliced the same regardless.
PavelH wrote:
Sun Oct 01, 2023 4:37 pm
3. Does it make sense to try designing an intersection with 4 lanes on one side and 2 lanes on the rest? Or am I unlikely to get much better than joining 4 lanes into 2 before the intersection?
The use case I imagine for this would be a 4 lane "spine" network, with several 2 lane "local" networks connected to the spine.
Since all of the large designs parallelize each path, I tried adapting some of them (namely "parallel multicross" and "parallel multicross turbo") to 4 lanes on one side, but the benchmark didn't really show any difference in throughput. I'm not sure whether this is because my design is bad or because the train direction mix used in the benchmark doesn't really match where such an intersection would be used. Any thoughts?
I've seen intersections that have 4 lanes horizontally and 2 lanes vertically, for example. Your adaptation was flawed. You would need to adapt the Parallel Multicross 4 Lane instead, by reducing the 3 sides from 4 lanes to 2 lanes. Then you'd probably see some throughput increase over a 2 lane Parallel Multicross. Personally, I would just have 2 <-> 4 lane conversions and leave the intersections alone. Intersections are the major bottlenecks in a rail system. Don't have merges/splits before/within/after them if at all possible. In reality, 4 lanes are rarely needed anyhow. Better rail system design is much better at preventing congestion.

Once again, I'd suggest going to the discord server. They're very helpful and friendly!
Avona

PavelH
Manual Inserter
Posts: 4
Joined: Sun Oct 01, 2023 3:33 pm
Contact:

### Re: 3 and 4 way intersections

Avona wrote:
Sun Oct 01, 2023 6:13 pm
I have not seen any advanced intersection writeup. Most tutorials go over the signals. I'd suggest asking questions in the Factorio discord server in the #train-help channel. There are many experienced people there who can answer many of your questions. That being said, I will go over the rest of your post.
Thanks!

Personally, I find forums easier to use (due to searchability/discoverability), but if Discord is the generally preferred channel, I'm happy to use that instead.
Avona wrote:
Sun Oct 01, 2023 6:13 pm
In the Parallel Multicross Turbo, and a few of the others, they utilize high throughput merges. Regular merges get about 27 tpm for 2-4 trains. High throughput merges weave together traffic so that trains will merge at high speed instead of coming to a stop before merging. They can get about 33-34 tpm with this sort of merge.
Ah, cool, thanks!

Doing some handwavy math, does this mean that 2 lane intersections will always be limited to a score of ~100 with regular merges, and ~130 with high-speed merges?
Avona wrote:
Sun Oct 01, 2023 6:13 pm
You also said "having too dense signals causes throughout to drop". In general, the more dense the signals, the higher the throughput. After a crossing, the exit blocks must be the length of your longest trains to prevent deadlock.
I've seen this described as "inchworm effect" but I don't know how dense the signals would actually need to be for this to kick in.
Avona wrote:
Sun Oct 01, 2023 6:13 pm
You would need to adapt the Parallel Multicross 4 Lane instead, by reducing the 3 sides from 4 lanes to 2 lanes. Then you'd probably see some throughput increase over a 2 lane Parallel Multicross.
Hmm, that's an interesting idea.
Doing this in a naive way (just merging the exits/entrances of Parallel Multicross 4 lane), I get a score of approximately 100 in the benchmark, which is roughly in line with the best 2 lane intersections, just bigger. This can probably be improved my using smarter merging and avoiding unnecessary crossings.

Repeating my adapted arm of Parallel Multicross (2 lane) to build a full 4 lane intersection yields a score of ~145. This is an improvement over the 2 lane variant, but far from the native 4 lane designs, while not really being smaller (blueprint here). Maybe it would still be an improvement in the described use case?
Avona wrote:
Sun Oct 01, 2023 6:13 pm
Personally, I would just have 2 <-> 4 lane conversions and leave the intersections alone. Intersections are the major bottlenecks in a rail system. Don't have merges/splits before/within/after them if at all possible. In reality, 4 lanes are rarely needed anyhow. Better rail system design is much better at preventing congestion.
Makes sense, thanks!

Avona
Long Handed Inserter
Posts: 62
Joined: Sun Dec 26, 2021 8:50 am
Contact:

### Re: 3 and 4 way intersections

PavelH wrote:
Mon Oct 02, 2023 8:58 am
Doing some handwavy math, does this mean that 2 lane intersections will always be limited to a score of ~100 with regular merges, and ~130 with high-speed merges?
In practice, regular merges can get slightly higher because trains will not always come to a stop. The high score of 33-34 tpm is generally not advised because, when trains are stacked so close together, the ups cost increases dramatically due to trains' braking distance constantly touching the occupied block in front of them.

In the test bench, for testing we limit the input throughput to 29 tpm because it's slightly less than the max lane throughput with a signal every 1 wagon length, I believe (or was it two wagon lengths... hmm). When the input tpm meets or exceeds the max tpm, weird, unrealistic, and undesirable train behavior occurs. This is the same reason that we don't have 100% crossing turns any more for Set 3. The test can be gamed because of the unrealistic train behavior. We want sets 2 and 3 to test extreme adverse conditions, but not so much that it's not a useful metric.
PavelH wrote:
Mon Oct 02, 2023 8:58 am
I've seen this described as "inchworm effect" but I don't know how dense the signals would actually need to be for this to kick in.
I searched with google and only found two references to this. I see some incorrect information that needs clarification. On a lane with no crossings, densely packed signals have no negative effect on throughput or ups. In fact, in order to reach the highest throughput on a lane, they will need to be as densely packed as possible. This changes with crossings. It's a similar concept to the high throughput merge. You can space out trains so that they reach a crossing just as the crossing traffic has just passed, so that the spaced out trains cross at a higher speed. So, it's helpful for certain intersections to have longer blocks before its crossings. It's easy to see this effect in some of the buffered intersections. I made a gif back in the day for the Hurricane's synergistic Set 3. Note that this was before we implemented the 29 tpm input max and 90% crossing/5% straights/5% non-crossing set 3.

SaiMoen
Burner Inserter
Posts: 10
Joined: Sat Nov 09, 2019 3:51 pm
Contact:

### Re: 3 and 4 way intersections

Since the train fff's came out I have experimented again and made this, could be a good one, haven't found it on here yet.

Illiander42
Filter Inserter
Posts: 512
Joined: Mon Feb 05, 2018 10:01 am
Contact:

### Re: 3 and 4 way intersections

Would someone be willing to point me at an explination of how "high-throughput merges" work, and how they're different from regular merges?

hansjoachim
Fast Inserter
Posts: 244
Joined: Wed Apr 26, 2017 7:03 pm
Contact:

### Re: 3 and 4 way intersections

Illiander42 wrote:
Fri Oct 06, 2023 7:43 pm
Would someone be willing to point me at an explination of how "high-throughput merges" work, and how they're different from regular merges?
It's not recommend because of the UPS cost. It's just either starting trains further back so that they merge with higher speed or combinators that makes an trafficlight system letting say 6 trains pass per lane.

sparr
Smart Inserter
Posts: 1337
Joined: Fri Feb 14, 2014 5:52 pm
Contact:

### Re: 3 and 4 way intersections

I know this is shouting into the void, but I'm already subscribed so I'll keep getting the reply notifications either way...

It would be considerate to use new threads in the appropriate subforums to ask general train questions, rather than using this single thread for every train related topic.

hansjoachim
Fast Inserter
Posts: 244
Joined: Wed Apr 26, 2017 7:03 pm
Contact:

### Re: 3 and 4 way intersections

sparr wrote:
Mon Oct 23, 2023 6:12 pm
I know this is shouting into the void, but I'm already subscribed so I'll keep getting the reply notifications either way...

It would be considerate to use new threads in the appropriate subforums to ask general train questions, rather than using this single thread for every train related topic.
Since ramps are coming this topic is dead. There are no reason to design any intersections again, just these simple ramps and mergers. Ramps have no speed reductions, are relatively small and will probably cost basically nothing. If they were long enough to lift a train and had a big speed penalty we could still see many intersections, but that won't happen.

Avona
Long Handed Inserter
Posts: 62
Joined: Sun Dec 26, 2021 8:50 am
Contact:

### Re: 3 and 4 way intersections

hansjoachim wrote:
Mon Oct 23, 2023 8:17 pm
There are no reason to design any intersections again, just these simple ramps and mergers.
There are non-ramp intersections possible that are more compact. Ramp tech comes later, so there will still be a place for non-ramp intersections. Even though there are fewer possibilities because of ramps, there will still be intersections to test. Rail curves have been redesigned, so the intersections either new or redesigned will all need to be tested.

I could see this thread being saved for a record of the past, and making a new thread for the new designs.

sparr
Smart Inserter
Posts: 1337
Joined: Fri Feb 14, 2014 5:52 pm
Contact:

### Re: 3 and 4 way intersections

hansjoachim wrote:
Mon Oct 23, 2023 8:17 pm
Since ramps are coming this topic is dead. There are no reason to design any intersections again, just these simple ramps and mergers.
That's not true at all. Ramps are huge. Just a ramp up, over one rail, and back down is 34 tiles long. The smallest useful intersection you can make with ramps is about 50x48 tiles, and that's just a diamond with no left turns. For a complete 2 lane intersection with ramps eliminating every crossing, you're looking at something like 100x100 tiles.

A) 100x100 is huge compared to the intersections in this thread that fit in 30x30 or 40x40. There's still plenty of reasons to use those smaller intersections. And with the new curves and diagonals, we have to start mostly from scratch on optimizing them!

B) At 100x100, there are a lot of different possible designs. It will probably be months before someone finds the smallest possible complete ramped 2 lane 4-way intersection, and then there will be different solutions if you want a square or diamond or plus shaped footprint, and for different rail spacings.

hansjoachim
Fast Inserter
Posts: 244
Joined: Wed Apr 26, 2017 7:03 pm
Contact:

### Re: 3 and 4 way intersections

sparr wrote:
Fri Nov 03, 2023 1:44 am
hansjoachim wrote:
Mon Oct 23, 2023 8:17 pm
Since ramps are coming this topic is dead. There are no reason to design any intersections again, just these simple ramps and mergers.
That's not true at all. Ramps are huge. Just a ramp up, over one rail, and back down is 34 tiles long. The smallest useful intersection you can make with ramps is about 50x48 tiles, and that's just a diamond with no left turns. For a complete 2 lane intersection with ramps eliminating every crossing, you're looking at something like 100x100 tiles.

A) 100x100 is huge compared to the intersections in this thread that fit in 30x30 or 40x40. There's still plenty of reasons to use those smaller intersections. And with the new curves and diagonals, we have to start mostly from scratch on optimizing them!

B) At 100x100, there are a lot of different possible designs. It will probably be months before someone finds the smallest possible complete ramped 2 lane 4-way intersection, and then there will be different solutions if you want a square or diamond or plus shaped footprint, and for different rail spacings.
.

- Ramps are not huge compared to the intersections in this thread. You are not taking into account the safe outputs that are needed for each direction. For example, a 32x32 intersection also needs 7 tiles per component (minus 1) for each exit, which makes it 73x73 for 2-4 trains. Ramps do not need such outputs, so they are more compact. With longer trains, ramp based intersections get even smaller compared to one plane intersections. For example, a 2-8 train intersection would need 101x101 tiles for safe outputs, while ramps would only need 50x48 tiles.

Since the throughput of ramps is much higher than regular intersections you do not need grids or alternative routes for trains, which allows you to save even more space.

The ramps in Factorio are unrealistic and unbalanced. They are too short for the steep angle they create. Assuming the trains are 3 m tall and need 1 m of clearance, the ramps would lift them by 4 m over a distance of 16 m. That’s a 25% gradient, which is much higher than the steepest adhesion railways in the world. Moreover, the trains don’t seem to lose any speed on the ramps making anything else obsolete. Sure there is some room for some roundabouts with your less than 100 trains before you get ramps.
If the ramps had a major speed decrease we could still have one plane intersections shine in some scenarios and we would have some special intersection design to overcome the speed decrease.
With these overpowered ramps there are no challenge left in designing intersections.

mmmPI
Smart Inserter
Posts: 3100
Joined: Mon Jun 20, 2016 6:10 pm
Contact:

### Re: 3 and 4 way intersections

Regarding the challenge of maximizing the throughput i agree that this is going to be not as interesting if one can use ramp, as there is then only merger in the junctions which i expect ( but may be wrong ) would roughly yield similar results.

However to me the challenge of making compact junctions is still there. Ramps or not, there will also be more possibilities for junctions higher than 4 ways.

Those would rather be called interchange, but i'm pretty sure many people on the internet will still be looking for blueprints of junctions. Either cheap, either some you can level up to ramps when they are unlocked, some being for square grids, but also more exotic shapes, and when you have a square grid, now you can join two diagonals with ramps over your factory for some 5 6 7 or 8 way junctions.

It is not going to be same exact thread, but i suppose a "junctions" thread where people post theirs and maybe another one for discussion so that there is no back and forth on the blueprint one would still be a useful ressource.

ChakatStormCloud
Manual Inserter
Posts: 4
Joined: Tue Feb 07, 2023 1:35 am
Contact:

### Re: 3 and 4 way intersections

hansjoachim wrote:
Mon Oct 23, 2023 8:17 pm
Since ramps are coming this topic is dead.
Ramps aren't coming to the base game, they're exclusive to the DLC. On of the devs said as much on discord :/
So this topic is still plenty relevant, and likely will be for a very long time.

datarza
Long Handed Inserter
Posts: 90
Joined: Thu Sep 15, 2022 1:55 am
Contact:

### Re: 3 and 4 way intersections

SQUAREABOUT plus my invention, maybe it was invented before?
Screenshot 2023-12-03 005830.png (1.34 MiB) Viewed 2374 times

It can be also used as 4 or 6 ways intersection with some limitation for middle lines (has no u-turn to all directions, but it can be done on the next same intersection) - probably it is smallest 6-way intersection with U-turn and changing line feature.

It may not be an ideal intersection, but the "killer feature" is using it as a corner of cityblocks where we have two separated exits/enters in the same corner (one for trains, second for fuel supplies).
Screenshot 2023-12-05 142228.png (1.14 MiB) Viewed 2121 times
Last edited by datarza on Tue Dec 05, 2023 7:23 pm, edited 9 times in total.

ElderAxe
Fast Inserter
Posts: 134
Joined: Thu May 18, 2017 8:04 am
Contact:

### Re: 3 and 4 way intersections

datarza wrote:
Sun Dec 03, 2023 4:35 pm
SQUAREABOUT plus my invention, maybe it was invented before?

Screenshot 2023-12-03 005830.png

It can be also used as 4 or 6 ways intersection with some limitation for middle way (has no u-turn to all directions). Actually, it is not a perfect intersection, but is a good for using as a corner for cityblock with two exits/enters, one for trains, second for fuel supplies.

Screenshot 2023-12-03 113052.png

Straight lines on opposite directions are connected on the inner curve. So only one train can pass at a time. Also left turns cross each other so they will also block each other.

datarza
Long Handed Inserter
Posts: 90
Joined: Thu Sep 15, 2022 1:55 am
Contact:

### Re: 3 and 4 way intersections

ElderAxe wrote:
Mon Dec 04, 2023 12:31 am
Straight lines on opposite directions are connected on the inner curve. So only one train can pass at a time. Also left turns cross each other so they will also block each other.
Actually it is not try, two trains can pass opposite directions:
Screenshot 2023-12-03 193812.png (3.54 MiB) Viewed 2339 times
Also, left turns do not block intersection:
Screenshot 2023-12-03 194705.png (3.78 MiB) Viewed 2337 times

Avona
Long Handed Inserter
Posts: 62
Joined: Sun Dec 26, 2021 8:50 am
Contact:

### Re: 3 and 4 way intersections

datarza wrote:
Mon Dec 04, 2023 12:40 am
Also, left turns do not block intersection:
They block each other:

datarza
Long Handed Inserter
Posts: 90
Joined: Thu Sep 15, 2022 1:55 am
Contact:

### Re: 3 and 4 way intersections

Avona wrote:
Mon Dec 04, 2023 6:24 am
datarza wrote:
Mon Dec 04, 2023 12:40 am
Also, left turns do not block intersection:
They block each other:
This situation is imposible, train can not move to top from bottom-right line as well as can not move to bottom from top-left line.

Probably you mean the mirror situation, but here there is not differences in comparing to SQUAREABOUT (it is not an issue for this intersection):
Screenshot 2023-12-04 084245.png (3.23 MiB) Viewed 2273 times

Avona
Long Handed Inserter
Posts: 62
Joined: Sun Dec 26, 2021 8:50 am
Contact:

### Re: 3 and 4 way intersections

datarza wrote:
Mon Dec 04, 2023 1:29 pm
This situation is imposible, train can not move to top from bottom-right line as well as can not move to bottom from top-left line.

Probably you mean the mirror situation, but here there is not differences in comparing to SQUAREABOUT (it is not an issue for this intersection):
Yes, I put the arrows backwards because I am used to the opposite RHT. The problem is still present.

datarza
Long Handed Inserter
Posts: 90
Joined: Thu Sep 15, 2022 1:55 am
Contact:

### Re: 3 and 4 way intersections

Avona wrote:
Mon Dec 04, 2023 8:44 pm
datarza wrote:
Mon Dec 04, 2023 1:29 pm
This situation is imposible, train can not move to top from bottom-right line as well as can not move to bottom from top-left line.

Probably you mean the mirror situation, but here there is not differences in comparing to SQUAREABOUT (it is not an issue for this intersection):
Yes, I put the arrows backwards because I am used to the opposite RHT. The problem is still present.
Sorry, can not understand the problem, please explain in details. I can confirm, that turning to the right from top (to the left in train's point of view) blocks the train, which turns to left from bottom (to the left in train's point of view). It is expected behaviour for unbuferred intersections (and for buffered probably as well) because they should cross the railway of each other.
Last edited by datarza on Tue Dec 05, 2023 1:29 am, edited 2 times in total.