Factorio Forums

I often find that if I unload goods into a bot network, the bots prefer provider chests in one direction or the other relative to the train. They will pick up all the goods from the provider chest closest to the front, for example, and the chests in the back fill up. The result is that trains get stuck even though the unloading platform theoretically has capacity, as one of the cars will not fully unload until the bots start grabbing goods from the back. In a high throughput situation, this doesn't work very well.

One solution I've found is to unload into one set of chests, then have belts where each chest is merged with the corresponding chest from the next car. Here is an example of what I mean, about to be employed in a purple base which was recently upgraded from belts to bots:

https://imgur.com/pXoJStX

There are a couple issues here. The more complicated a balancer used, the worse the UPS hit. For example, instead of having 3 splitters in a row, I could use a 6 to 6 balancer for even better unloading, but then there's a lot of physical space used and a lot of merged transport lines. Another issue is that it doesn't really scale well to three or more cars, unless I like having giant spaghetti belts coming out of my chests.

Any suggestions for a better solution? The desired traits are to have the cars unload in a balanced manner, as fast as possible, without using enormous amounts of space or killing my UPS.

One option would be to have provider chests feed into buffer chests, but then the work being done by the bot network becomes very spiky. A train shows up, and a thousand bots converge on the chests to redistribute the goods, then go charge for a few seconds while the rest of the base sits idle because no bots are available. Probably fine at the home base where the bot network already has tens of thousands of bots, but not great in a satellite where there are only a couple hundred bots.

Well... try this. unloaded evenly for me wherever the bots came from. (latest 16 experimental, tested in creative mode)
Test setup:
Combinator and lights were for testing, not included in blueprint.

Certainly looks promising, the only issue is the spikes in bot usage when a train arrives. I worry it would make it infeasible in a small outpost. The other potential issue would be if you have delivery stations on opposite sides of a large base - now the bots might decide to bring an iron plate all the way across the base, for example.

Thanks!

Bots will always grab from the nearest chest. You could solve it by using fewer chests, unloading into 2 iron chests, and have both of those unload into a single active provider. Works for a 2 wagon train only though.
Since you can't turn logistic chests on and off, there is no way to tell bots to grab from specific chests.

My preferred solution, at least in large bases, is to set up a transfer area for bulk cargo, on a separate logistics grid, then use something similar to Disentus' method to transfer cargo between trains coming from mines to storage to bigger trains leading to the production sites. This keeps traffic on my base rail grid down to acceptable levels (all mining traffic is pushed to a separate railgrid on the outer perimiter to the transfer areas on the south side of the base - meaning mining trains never go through the production area. A major traffic saver)

For your own system, it kinda depends on the needed throughput. Personally, I'd just dump the output into a single red chest as best as able, but that'd bottleneck you by at most 4 stack inserters used. Using passive provider chests like this is a pain in the neck to do balancing with. The only way I know of is to just set the requestor chests to a high value so that they act as a buffer for whatever production needs the trained-in resource.