This is the simplest I'd do:
-
Have a dedicated refueling line. This one is a bit special and probably with a 1-1 train it'll be enough. This line should go from "fuel pickup -> fuel drop off". Maybe, down the line, more than one refueling train will be needed. Your fuel drop off should refuel this line first. This way, as you progress you just need to change what's in "fuel pickup" and all your trains will (eventually) use your new fuel.
-
Have a* refueling area where all your trains are refueled. Think of having 5-10 (as many as you need) stations in parallel, all canned exactly the same (i.e "Refuel") *These stations might become a traffic bottleneck so you might want to have a few returning areas scattered around. For automatic schedule you can use a few seconds of inactivity as a condition.
-
Have a parking area. Think of 10 stations all named exactly the same (i.e. "Parking"). These might also become a traffic bottleneck so you might want to have a few of these areas scattered around. Use no condition in the schedule, so trains won't stop unless they have nowhere to go. It's important to consider these areas, as you don't want a train which can't pickup anything blocking a refueling station.
-
Name every pickup station type exactly the same (i.e. 1-4 Iron Ore Pickup) Use amount of material in the station as a condition to set train limit (and not enable/disable station). You can use circuits to set exactly the number of trains you can serve but I usually don't bother. If you can request 3 trains, you need to make sure that there's space to park 2 trains by the station where they won't interrupt traffic. I did it simpler to just have multiple pickup stations in parallel and depending on waiting material request 1 or 0 trains. As multiple trains would be waiting anyway, I prefer them being loaded. This might not be the most efficient way but I think it's simpler. For the scheduler you'd use full cargo OR Time passed. If a couple of minutes have gone by, have the train depart and not wait anymore.
-
Name every drop of station exactly the same (following the iron ore example, 1-4 Iron Ore Drop Off) As schedule conditions you'd use Empty Cargo OR Time passed. If a couple minutes go by, have the train depart and not wait anymore. You'd use circuits to keep requesting trains until the station has stopped as many material as you want. Keep in mind that if you request more than one train, you'll need space for the remaining trains to wait without interrupting traffic.
Your train schedule should look like (adjust times add needed):
- Refuel: Inactivity > 5 sec
- Parking: No conditions
- 1-4 Resource Pickup: Full Cargo OR Time Passed > 60
- 1-4 Resource Drop Off: Empty Cargo OR Time passed > 60
You can repeat parking, pickup, drop off a few times to minimize refueling (be careful not to do this too many times so your trains spend all their fuel before refueling) If you do, remember that after drop off you should always go to parking.
Once you have the schedule set up for a train, you can copy it with shift+right click and paste with shift+left click.
Finally, it's important to use train limit instead of enable / disable.
Using the former, over a train has departed, if the station changes and requests 0 trains, the train will keep going.
If you use disable, if a train departs and the station becomes disabled, the train will stop in the middle of nowhere, potentially causing a gridlock.
PS: I usually name my stations with the type of train, so if later I decide to bring ore from an outpost far away, I can make a few exchanging hubs with 3-8, 4-16, ... bringing materials and 1-4 distributing then around the base
I start with a bus leaving enough headroom for multiple stations in the top part and a mall close by.
After trains, first I bring ores and then I start building minifactories for dedicated production drifting away from main bus.
When bots are available I start by converting the mall and then the main base is turn more or less into city blocks, but keeping the main bus.
Not as extreme as this but you can get the idea.
Fot the moment, it has worked everytime (including K2, let's see how it goes with current run of B&A) and the main plus (IMHO) is that I don't end up with a base that looks the same every run.