Train based direct insertion challenge

[Amarula]

A recent post on reddit caught my eye with a comment about common items that can hurt your UPS as your base grows into the mega-base range: belts and bots. Now you have to have some means of moving items around, and for a large base that typically already means using trains to bring in raw materials from your outposts. So I started thinking what if I used trains for everything?
1. No belts. Everything is transported by train. *
2. No bots. Everything is transported by train. **
3. No chests. The only storage capacity is what is being carried on the train. ***
4. Direct insertion to and from trains only. ****

*, **, ***, ****: A few exceptions are allowed for low volume, namely satellites for rocket launches, kovarex, nuclear fuel cells for powering the reactors, and nuclear fuel for powering the trains, all of which are being delivered by bots.

So I am transforming my plain train based city block to meet this challenge. No it isn't particularly efficient but it is fun

To start, all the miners are going to mine directly into trains. Ore is taken to the local furnace and turned into plates. You will see in the first photo below, ore trains feed two rows of furnaces, and the furnaces feed the delivery train from both sides. This is a tiny outpost so there are only two trains for copper plates, one loading, and room for one to wait to load. For larger outposts there will be stackers to hold the waiting plate trains. You will also notice that the ore trains are on their own dedicated tracks, there is no way for them to get on to the main train network.

No Belts mining outpost.png (4.08 MiB) Viewed 14612 times

Horizontal and vertical tracks are always on an even boundary, so you can either load or unload with a fast stack inserter, but the other side has to use a long-handled inserter. For furnace and factory output this isn't too much of an issue, and you can add multiple output inserters if one can't keep up. But for transferring plates, I wanted something able to handle more throughput...

[Amarula]

Now a train station has be placed on a straight piece of track, and two side by side straight tracks are two spaces apart, so you can't use fast stack inserters. But there isn't anything to prevent the rest of the train being on a diagonal track, and diagonal tracks can be a single space apart. So I am proud to introduce a train transfer station that I call... Walk Like an Egyptian!

No Belts Walk Like an Egyptian.png (1.15 MiB) Viewed 14608 times

You can only squeeze in four stack inserters per wagon, but that is still much faster than using long-handled inserters. There is also room to a add a third track, so the middle train can be loaded from both sides at once...

[Amarula]

When using trains to deliver direct to the factory, the train has to have the right ingredients for the factory. Now I could just add ingredients and leave whenever I run out of any of them, but (is my OCD showing) I am choosing to set the trains to have the exact amounts so that everything runs out at the same time. For items like gears that have a single input this is trivial; for items like red science, where the items have the same amount and the same stack size it is easy.

No Belts Red SP factory.png (1.73 MiB) Viewed 14608 times

[Amarula]

For items with different amounts and different stack sizes, it gets to be more challenging. Green circuits aren't too bad, 2 stacks of iron at 100 per stack for every three stacks of cables at 200 per stack. But what about science packs for the research labs?
Space science at 2000 per stack! Six other science packs at 200 per stack. I can fit 1K of each colour using 5 stacks of 200 times six colours, but I plan to limit space science to half a stack. I can do this by only loading one rocket launch per wagon which conveniently happens to produce 1K space science.

No Belts research park.png (1.85 MiB) Viewed 14592 times

One nice thing about the labs, there is no output to put on a train, so I can feed labs from both sides of the train, and fit it in some beacons to keep my research speed up...

I have created a staggered start for the research trains. The park has 5 rows of trains. The first one started with only 1 stack of potions, the second with two and so on. This balances the demand for science production as each lane empties the train in sequence.

[Amarula]

I am using both filtering the cargo wagon slots and limiting the wagon size as needed.

No Belts research wagon contents.png (110.93 KiB) Viewed 14591 times

[Amarula]

I haven't actually got all my science converted yet, so I have yet to run any research with this new challenge. So far I have red and green, and I am working on blue. This will bring on the challenges of dealing with oil production and delivering liquid ingredients... the factory must grow!

[Koub]

Walk Like an Egyptian!

Is this a Jojo reference ?

Back to topic

items that can hurt your UPS as your base grows into the mega-base range: belts and bots

afaik, bots are not an issue for UPS. That's why as megabase grows, they tend to replace every other solution : because they're so damn UPS efficient.

[eradicator]

items that can hurt your UPS as your base grows into the mega-base range: belts and bots

afaik, bots are not an issue for UPS. That's why as megabase grows, they tend to replace every other solution : because they're so damn UPS efficient.

As far as i know (and that's not very far) belts were better than bots after the last belt-ups-optimization.
Here's an old thread with a belt only base.

[Amarula]

items that can hurt your UPS as your base grows into the mega-base range: belts and bots

afaik, bots are not an issue for UPS.

Sorry I missed the other part of the comment which was about needing a lot of power, which included beacons and bots.

[Amarula]

I came up with a simple calculator to figure out how many stacks of each input I need for a perfectly balanced train.

Input train content calculator.png (17.94 KiB) Viewed 14251 times

I haven't added a database lookup so I just enter the inputs, and the stack size and number required for each input.
Then I take a guess at how many stacks of the first item will fit; this is used to calculate the total number of outputs made from that many stacks, which in turn is used to calculate how many stacks of the other inputs are needed.
Then all the stacks are added up so I can make sure I haven't exceeded the limit of 40 stacks per cargo wagon. I adjust the initial guess until everything fits.
Finally, it calculates how many of each item will be on the train, which I use to set the controls for leaving each station in the train schedule.

[Amarula]

My current sector design for up to five rows of factories:

No Belts Red Science map view.png (60.61 KiB) Viewed 14249 times

Consumers enter at top left, with a stacker for up to five trains. Note the queue station to force trains to repath before choosing which lane to enter; I tried a few different designs but this was the easiest that (so far) works every time.

Suppliers enter at bottom right, with their own stacker for up to five trains. Again there is a queue station to force repathing.

Consumers cross the supplier exit lane in order to enter their lane, and suppliers cross the consumer exit lane likewise. I don't see any simple way to avoid that while still having a single stacker.

One issue with the train only part of the challenge: the factories only work when they have a supplier train in the station providing inputs. So it does no good for a consumer to enter a 'free' lane, if it doesn't have a supplier in the station. So I have added a control wire from the supplier station to the signal, to force consumer trains to choose a lane with a supplier. It slows things done a touch, but the whole thing is so slow anyway (how long does it take two factories to fill a cargo wagon with gears?!) I think this is acceptable.

No Belts station control signal.png (185.94 KiB) Viewed 14249 times

There are sidings East and West for personal transport train to visit the sector; the siding at the bottom is for the fuel train that delivers train fuel, which is delivered to the supplier trains via bots (one of the allowable exceptions for this no belt no bot challenge).

[Amarula]

A thinking out loud post, hoping for some brilliant insights: So in my mind I have this picture of a perfect world (lol but of course it is in Factorio) where the factory sector is working full out: there are five consumer trains loading the sector outputs, and five supplier trains unloading supplies. There are five more supplier trains out loading up with the inputs for this sector, and at least one of those is always full up and back in the supplier queue by the time before it is needed.

I think this is going to work okay for relatively simple sectors like red science; gears are produced much faster than the red science factories use them, and copper is picked up directly from waiting supply trains in the copper ore WH that stacks all the trains coming in from the supply outposts.

But what will happen for sectors like blue science? The supply trains need to be at the engine sector (or the engine factories stop working); they also need to be at the red circuit sector or those factories stop; and they need to be picking up sulfur. I am afraid that five loading trains aren't going to be able to keep up. How to calculate how many trains I need as opposed to what needs to be on each one? 20 trains is worst case, but do I need that many? Because I don't have room in the stacker to hold fifteen trains (5 each for loading engines, red circuits, and sulfur). I can push the stacker to hold ten trains; when everything is working, most of the trains will be out loading supplies, but what happens when all 15 want to queue at their home sector?

[mrvn]

You could try using bidirectional trains with one cargo wagon at the end like this: <>C. A forward locomotive, a backward locomotive and a cargo wagon. The station design then has to be T shaped with an extra stop (makes a nice stacker). The train enters from the bottom and drives to the extra stop. It then reverses direction and slowly drives backwards to the real stop (wagons have a high wind resistance). You do your loading/unloading and the train drives out in forward direction. Now what would be the point of this? You can have 4 train tracks per assembler, one on each side. All 4 sides with stack inserters. No need to get mixed wagon loads unless you need more than 3 items to craft one result.

Another approach might be to use the same train for input and output. So load a train up with iron ore and copper ore. Drive to the iron smelter and switch iron ore for iron plates. Drive to the copper smelter and replace copper ore for copper plates. Drive to the copper cable assembler and replace copper plates with copper wires. Drive to the electronic circuit boards assembler and replace iron plates + copper cable with electronic circuit boards. And so on.

Also have you tried placing assemblers between diagonal tracks?

[Amarula]

You could try using bidirectional trains with one cargo wagon at the end like this: <>C. A forward locomotive, a backward locomotive and a cargo wagon. The station design then has to be T shaped with an extra stop (makes a nice stacker).

Cool idea!

You can have 4 train tracks per assembler, one on each side. All 4 sides with stack inserters. No need to get mixed wagon loads unless you need more than 3 items to craft one result.

I am not sure how this would work - one space for inserter, 3x3 for factory, plus one for inserter is 5, but train tracks have to fall on even boundaries, so at least two sides would need long-handled inserters - not an issue for output, but one of the inputs is going to be much slower than the others.
Also you can only be using two sides at a time, either vertical or horizontal; the trains on those tracks would be blocking the other two lanes. You may have missed the 'no chests' part of my challenge, I am unloading from the train direct to the factory without allowing any chests in between.

It is still a cool idea!

Also have you tried placing assemblers between diagonal tracks?

No haven't tried that... I saw a recent post over on reddit about building on the diagonal to allow loading from belts (would also work with chests) from all four sides, was interesting food for thought...

[mrvn]

I was thinking like this:

4way.png (2.4 MiB) Viewed 14204 times

But that won't work for assembler.

There is a hack where you place a cargo wagon on 2 rails and then you can place inserter under the wagon because it extends over the end of the rail. But with train stops the train doesn't even drive far enough to put inserter at the end of the rail. You can only load/unload from the sides and then I see no way to place 4 stops within inserter reach.

Still you can do 2 stops with stack inserters and 2 with long inserters. Double up on the long inserter for extra speed (I think you might fit 6 per stop).

[Amarula]

I was thinking like this:

Mind blown away!!!!

[Stevetrov]

DI train builds are currently my favourite factories to design, they can be really UPS efficient and they provide a different challenge to belt based factories.

3. No chests. The only storage capacity is what is being carried on the train. ***

This restriction is probably going to hurt you a lot, to maximize UPS efficiency you want to use max prod modules and a decent number of speed beacons. With this constraint the max you can do is 4 beacons per ASM


I would relax this restriction to allow yourself upto 1 chest between machines / machine & train.

EDIT: I could be wrong about this because we know that inserter cost is relatively high in UPS so minimizing inserter swings is important, but is it worth sacrificing that many beacons? maybe in some cases but I dont think it is in most.

Here are a couple of links to DI train bases that I have built.

https://www.reddit.com/r/technicalfacto ... wo_editor/

https://www.reddit.com/r/technicalfacto ... map_style/

[DaveMcW]

With this constraint the max you can do is 4 beacons per ASM

You can use 8-tile spacing and 2 long-handed inserters to get 6 beacons.

[Amarula]

So I finally made it to what I hope is the last challenge of designing this direct insertion train base: the rocket silo!

My research trains are balanced for each wagon to hold 1K of each research pack, so five stacks of 200 for each of red, green, military, blue, purple, and yellow. And one half stack of space

However, 1K works out quite nicely to be one rocket launch per wagon. So my silo loading site has multiple train stops, and the train stops once at each to load the next wagon. The inserter for the satellite is enabled only when there is a train at one of the stops, so the train stops, the satellite is inserted, the rocket launches, the wagon gets loaded with 1K of space science, and the train moves to the next stop. It takes long enough to launch the rocket, that there is time for the train to move to the next stop, so I don't accidentally get more than 1K loaded into any of the wagons.

The train stops don't align perfectly, so every other wagon only gets five inserters doing the loading, but it still works acceptably quickly. And the entire design works for unloading the rocket part inputs, simply changing to long handled inserters. Again, the long handled inserters are slower, but they work quickly enough to launch a rocket every 3 minutes.

Now I am thinking about doing it over (because what else do we do in Factorio ) to put the train stops on the outside of the tracks instead of the inside so I can place multiple rocket silos and launch even more rockets

I put it down and it worked perfectly the first time through, and holy space science Batman but that felt good!

No Belts rocket silo.png (2.33 MiB) Viewed 13660 times

[Amarula]

New improved direct insertion rocket silos make me happy!

No belts improved silo.png (4.21 MiB) Viewed 13444 times