Launch Debrief

After hours of blueprinting and spreadsheeting, I’ve finally launched my first rocket in Factorio. I was rushing to get it done, and it happened just after 2:00 in the morning, so there are a lot of notes about the experience I need to get out of my head.

My special goals, in addition to launching a rocket, were to:

It’s taken a few days to write this properly, and I’ve been playing in the meantime. I’ve got my nuclear build running enough to supply power for now, played with the spidertron, and am closing in on completing all non-infinite science.

T-Bus

I’m unreasonably satisfied with the squaricity. However, I noticed that the T-junction was tricky, and ended up noodlier than I’d like. The two output busses should have their items permuted relative to each other so that the t-junction can be built of branch-offs. That said, here’s a wonderful 4-to-2+2 balancer.

I noticed that commonly-used-together items should not be adjacent on the bus if they each have only one lane. It’s definitely possible to take a half-half belt in these cases, but I find it a bit ugly (though I’ve yet to find a non-ugly branch-off anyway). It seemed reasonable at the time to organize this bus to have the most-commonly-used things closer to one side, but in retrospect the cost of having something on the other side of the bus is practically nil.

I tried a bit of gapless bus, but I think I need more experience using it to make it more enjoyable. In particular, I need to commit to it, whereas this time I kinda tried gapless as an afterthought. If I can avoid the annoyance of finding gaps

Blue Jump Start

My starter ore patch got me through nearly all of red, green, black, and blue science. It probably could have gotten me all the way if I had used my ore patches more efficiently (but I decided to focus on expansion before they ran out).

I built it on-the-fly, which is good and bad. The good part is that the items are created in the order in which they are actually needed to bootstrap itself. The bad part is that there’s rather some spaghetti that could have been avoided with some planning.

Here are some things I think could use better design:

Mega Multitool

A while back, someone posted their rediscovery of a 4-asm multitool accessible immediately after Automation 1 tech. In the comments, someone mentioned using a cargo wagon and circuit logic to beef it up. Their circuit idea was modded-only, but it got me thinking. So I designed a multitool that takes iron plates, copper plates, iron ore, stone, coal, water, and crude, and produces all sorts of stuff up through using blue circuits and LDS: just about anything except nuclear.

I didn’t set it up early-game enough, it probably needs to be part of the jump-start base. Essentially, I’d like it to construct my first bots for me, but that means building it without bots. As such, I think I should simplify it somewhat.

In retrospect, producing lubricant from crude in the tool was probably not a good choice. Instead, if the jump start has oil processing and plastic production in it, that simplifies a lot of the design. In fact, I might split both red circuits and sulfur off from the blue science build, since the multitool can really start hurting for reds when making modules for personal equipment. I might keep sulfuric acid production on-site though, since early on it’s only for batteries.

What I was impressed by was how well it was able to bootstrap new processes. Before I had my production mall up, I could create centrifuges through the multitool. It also excelled at personal equipment: solar panels, batteries, lasers, shields, and so on. In fact, after the rocket launch, it created all sorts of intermediate components for my spidertron and power armor while I carried on scaling the factory.

Trains

Schedules for Mixed Trains

The plan was to have trains that carry several types of cargo. One thing that bit me in the interface was the inability to reorder wait conditions on trains or do anything more complex than sum-of-products logic.

Unloading multiple types of items at base is dangerous because the train might arrive with plenty of one resource but almost nothing of another. I ended up using conditions that looked like:

  OR
  ├ AND
  │ ├ iron ore = 0
  │ └ 2s inactivity
  ├ AND
  │ ├ iron plates = 0
  │ └ 2s inactivity
  └ AND
    ├ steel plates = 0
    └ 2s inactivity

Further, supplying fuel (or acid for uranium) to outposts was a bit tricky. It seems they could end up evening themselves out after a while, but I found the timescale to be too long generally. So I ended up tweaking amounts during train startup and turning them back after the outpost’s buffers filled. Choosing appropriate full-enough conditions to leave an outpost was another tweaking process but over a longer timescale as the factory grew. Large trains would end up waiting too long to fill completely even though my bus was a fraction of its final size. I ended up starting with a much lower fill requirement for copper plates and tweaked them upwards as the factory grew. One solution might be to use circuits to measure flow rate and depart if it’s too low on the outpost side, or hold departure if it’s too high on the input side.

The plan was to have a train that would do water+stone+uranium, and another at the same station for just water. I am expecting the separate water train to block an intersection (at the rail crossing gate). I’m not sure if I’ll change the water train to be a duplicate of my misc train, or make a separate water station (I even have a spare station set built from before I designed my train cargo layouts). In fact, I ended up running low on stone, and my centrifuges weren’t backed up, so I went ahead and made two full misc trains anyway.

Grand Central

As designed, the rail crossing gate requires a signal before it, which allows a train to pull forward from parking and block all other parking.

I think a tiny two-headed passenger train is a better choice than a long one-headed train passenger+rolling mall design. Such a train can be placed outside the normal rail parking.unloading area, and so doesn’t even need a rail-crossing (and neither would the loading/unloading warehouse area. That said, I do still need a rolling storefront: I ended up almost always making one or two trips back to base when expanding.

Specialized Trains

I’ve kept my expansions purposefully minimal because I don’t have a munitions/repair supply train. Future games might not be so light on enemies (I think I got lucky with my world seed), so I really need to put the effort in to create these small stations: they are just as complex as bulk goods stations. The one resupply addon I do have is for refueling, which doesn’t make much sense with a central station.

Train Unloading

I initially ignored the need for balancing, assuming MadZuri’s smart inserters on unload would handle it. Unfortunately, as soon as some belts back up to a wagon, the associated chests can’t empty properly, and the train will sit around with non-empty cargo and nowhere to put it. Since I designed the base with non-power-of-two belt widths, I resorted to some “good enough” balancers that haven’t caused a problem at the throughput I have so far.

In a design world, I’ve played around far too much with circuits to try to rebalance input weights based on detecting output backups, but to no avail (not scalably at least). Instead, I think I have to bow to research and just use the in-game mechanism: bot unloaders.

Since I had so much bot-less success elsewhere in the playthrough, I was really hoping for another bot-less approach here. However realistically you’ll only be pre-bot or post-bot state in a normal game for any length of time. If pre-bots, then your input requirements are low enough that you can use a small balancer, probably 4x4. If post-bots, then just use bots since it’s otherwise either design a massive balancer, or create a likely UPS-intensive system that simulates bot balance logic. And you know what, the potential of bots+trains is enormous; I quickly made a design and it unload six blue belts from a single wagon for an average of just under 6MW (okay, so it’s 0.07% less than a full six belt, I might even care…).

There is one botless option though: 1 wagon per train. There are definite advantages for trains being small: faster acceleration and breaking, smaller run-outs from intersections, simpler train schedule conditions, smaller stations, and the ability to see the entire length of a train when planning tracks. When 40 stacks of items isn’t enough, add more trains; when traffic gets congested, add another lane.

Circuit Conditions

One thing I’ve noticed about myself is that I don’t want to stop in the middle of a game to create circuits. Therefore I should really put the time in to create a book of circuit setups. In particular, I’m thinking:

Train Conclusions

This is by far the longest section, but ultimately it just means I need to try some very different approaches in station design. At least I now have an idea of the real problems with train to belt transfers, the early- vs. late-game inflection points, and my own habits.

75SPM

Up to launching the rocket, I only really hit 22.5, which I expected given that I’m only supplying yellow belts worth of copper to circuit fabs designed for blue belts. I suppose it’s no surprise that blue circuits were the bottleneck, thanks to insufficient copper for green circuits. I suppose it’s a good sign that so the design tracks known gameplay patterns.

I’ve gotten it to peak 75 and sustain 60+, but after the first rocket it’s mostly about supplying enough materials to the base as a whole. I ran into stone, steel, and multiple iron problems, though oddly circuits have been fine since the first launch.

Belt-based “Logistic System”

Worked like a charm! It was super fun watching dense sushi thin out as everything got sorted into place.

My only complaint is that you get way more items than you asked for of certain products. Green/red wire stands out to me, but I don’t think I care about that. On the other hand, I loaded bricks in directly from the bus, and it overwhelmed my flimsy inserter. Loading from bus should absolutely go through an inserter first, and probably a slow inserter.

I could experiment with different geometries to make it non-linear, but it fit very nicely next to my processing columns and doubled as a spare path. I could definitely do it again, but I’m interested in trying a train-based logistics system… or perhaps just using logistics bots ;) I might even make a walk-through make-everything as is traditional, but try to integrate better paths for human convenience, though I did like not having to exit my car.

Oh, one thing I wish it had was a recycling system. I’m overflowing with yellow belts now that I’m upgrading to red

Future Design Opportunities

I used 4-wide paths without thinking too much. It turned out too narrow for reliable car use without the vehicle snap mod. I should put some work into creating my own style of path and surrounding standards.

I should design a bus cross section that lets roboports and power poles reach across. Bonus points for lighting and obstacle-safe paths.

Having worked with some mixed trains, I want to push myself for a train bus base.

I don’t like my uranium processing column. Or my LDS column.

There are several oddball ingredients used in various places: I’d like to run them down a bus, but not each on their own lane. Instead, I should work with a sushi belt down the bus, or a bloodbus. If sushi, can I load new items at different points on the belt? Also, can I use one “sushi slot” for both crude and empty barrels?

Concrete is produced very much in bulk. As far as a mall (storefront) goes, it’s more like brick than anything else. Perhaps I should put some work into integrating concrete columns into brick columns.

I want a Kovarex setup to start all the machines as quickly as possible: i.e. no buffer in the machine. Addendum: I actually built one today which I quite like. Note that it takes 40 minutes to produce the 40 235U needed to kickoff another centrifuge, so the fastest way to bootstrap is with one machine until you’re satisfied with your stockpiles.

I used 4-wide paths just because I’m watching some Nilaus at the moment, but I think it’s too narrow for not having Vehicle Snap. In my first map, I planned for much wider paths, and perhaps I should go back to that and develop my style. There are a lot of considerations to balance

General Clean-up

These are specific notes about problems in my blueprints.