Page 1 of 1

Insert behavior

Posted: Wed Oct 17, 2018 2:58 am
by Mr. Tact
In my main base I have a large copper smelting array which has two train station inputs, and two station outputs. Here is a screenshot of the stations:

inserter behavior.jpg
inserter behavior.jpg (790.32 KiB) Viewed 2509 times

The station works reasonably well, but there is an oddity which I have a guess for, but I'd like a confirmation. The plate holding station chests do not seem to fill evenly. This is despite the fact the belt is never emptied, ie. there are always plates on the "back side" of the belt. So, none of the inserters should ever fail to get a full load. I am guessing, this is due to minor differences for the inserters which are pulling the plates off "curved" belts. Would you expect this to cause a significant difference in the amount in the plate holding chests even if you assume there is always a full bell of plates incoming?

Re: Insert behavior

Posted: Wed Oct 17, 2018 3:10 am
by DaveMcW
Yes, inserters have different swing times depending on the shape of the belt. Not that it matters in this case, all the chests are filling the same wagon.

Re: Insert behavior

Posted: Wed Oct 17, 2018 3:21 am
by Mr. Tact
Does it "matter"? All depends on the context -- but yeah, in 99.9% of contexts it is meaningless. However it annoys the heck out of me to see one chest full when the other is only half full. :mrgreen:

Re: Insert behavior

Posted: Wed Oct 17, 2018 4:11 am
by Zavian
Inserters also take more time to pick up if they need to switch lanes. (The switching lanes bit takes time).

If you really want to keep all the chests the same, then search google for MadZuri smart loader. eg https://www.youtube.com/watch?v=4IjfnOSHZ1A.

Personally I'm not sold on the benefits, unless you actually have a situation where the unbalanced chests are causing actual problems.

Re: Insert behavior

Posted: Wed Oct 17, 2018 8:55 am
by eradicator
The problem is that inserters grabbing from the same belt steal from each other. I.e. if two inserters swing towards the belt at the same time, then only the first inserter gets to grab items, while the second has to wait for new items to arrive.
This can be solved via circuitry ofc, either with measuring the chest content and enforcing an average content (like madzuri linked above) or via a clock that ensures inserters swing in the right order.
But the easiest way is really to give each inserters it's own lane:
fairshare.png
fairshare.png (331.32 KiB) Viewed 2470 times
Using circuits costs less UPS though.

Re: Insert behavior

Posted: Wed Oct 17, 2018 9:09 am
by Zavian
eradicator wrote:
Wed Oct 17, 2018 8:55 am
Using circuits costs less UPS though.
I think that just allowing the chests to be unbalanced would be even more UPS efficient. (Assuming that the unbalanced chests aren't causing actual problems).

Re: Insert behavior

Posted: Wed Oct 17, 2018 3:15 pm
by Mr. Tact
Yes, I am aware of the Madzuri smart loader. In fact, I use it at all of my ore outposts. However, there the issue was more a case of that without the smart loader plates don't get to all the inserters since the design is slightly different. In this case, the plates are always available to the inserters so I didn't think the smart loader "would be necessary", and at some level it isn't. It is just the standard Factorio OCDness which raise "the problem". :mrgreen:

And if I am going to do that, I might as well as also do something about disabling the first output station when the number of plates there differs too greatly from the second station. Since without doing that, the second station is only used when the first station is occupied. Leading to occasionally a train sitting in the first station waiting for plates to be delivered when the second station is nearly full.

Re: Insert behavior

Posted: Wed Oct 17, 2018 6:27 pm
by FredHp
Try this mod:

https://mods.factorio.com/mods/Atria/WideChests

Merging Chests. (This mod is so essential that should be integrated into de game!)

Just replace the chest with a merged one and your balancing problems is solved.