Strange belt spacing issue (possible bug?)

[SteelSlayer]

Recently I've been having this problem with my copper plate transport belt that I've never seen before. Essentially the inserters aren't able to place the copper plate into the tiny spaces you see on the belt. Meanwhile, the iron plate belt seems to not have this problem. This seemed to happen randomly when I came back to my game, loaded it up and started playing. Before, I was having no issue like this. Could this be a bug?

Factorio compression issue modified.jpg (1.01 MiB) Viewed 2800 times

Also heres another spacing Issue that has come up down my copper plate line that I can't explain.

20160817075924_1.jpg (734.88 KiB) Viewed 2791 times

[Frightning]

The first is a common problem: Inserters don't always manage to compress the belt, it's basically a matter of the timing of earlier insertions of plates causing the gaps between plates to be narrower than a single plate but still>0. One way to fix this is to find the inserter(s) that aren't being able to insert and make the belt under the tile they would insert on faster than the rest of the belt (note that this doesn't work w/ Express belts because they are the fastest possible belts!).

The second is an example of how the splitter behavior can be counter to intuition. The reason the first siding is only getting half throughput (did you notice that that belt is exactly half saturated?) is because of the splitter further down is having to put a full belt's worth of throughput out the latter siding. What's happening is this: The first splitter in the second set of 3 splitters will take 50-50 from incoming plates to feed the left exit (which is only one not plugged up). Following that fact backwards to the first set of splitters, we see that the last of said splitters is thus having to supply 1 belt's worth of throughput, which it will take 50-50 from inputs. This is key because that means the first splitter of the first set is going to put half a saturated belt worth of throughput onto the bottom output (which feeds into the last splitter of first set of three). Now we look at the middle splitter, it will attempt to split it's input 50-50 onto output, so half a belt worth at the third splitter, and half a belt worth at siding, and it succeeds at both. Notice what's happened: Despite the fact that we are supplying 2 belts worth of throughput to the balancer, it ends up finding an equilibrium that only assigns half a belt to the first siding and a full belt to the second.

[Zeblote]

The first looks like a bug, inserters were recently updated to fix exactly that.

Why don't you post the save file?

[SteelSlayer]

Here is the save file. I'm using some mods like bob's mods etc but the smelting setup is just a bunch of vanilla stone smelters and inserters so it should be fine.

Bobs_RSO_Marathon.zip

(8.02 MiB) Downloaded 100 times

[SteelSlayer]

The first is a common problem: Inserters don't always manage to compress the belt, it's basically a matter of the timing of earlier insertions of plates causing the gaps between plates to be narrower than a single plate but still>0. One way to fix this is to find the inserter(s) that aren't being able to insert and make the belt under the tile they would insert on faster than the rest of the belt (note that this doesn't work w/ Express belts because they are the fastest possible belts!).

The second is an example of how the splitter behavior can be counter to intuition. The reason the first siding is only getting half throughput (did you notice that that belt is exactly half saturated?) is because of the splitter further down is having to put a full belt's worth of throughput out the latter siding. What's happening is this: The first splitter in the second set of 3 splitters will take 50-50 from incoming plates to feed the left exit (which is only one not plugged up). Following that fact backwards to the first set of splitters, we see that the last of said splitters is thus having to supply 1 belt's worth of throughput, which it will take 50-50 from inputs. This is key because that means the first splitter of the first set is going to put half a saturated belt worth of throughput onto the bottom output (which feeds into the last splitter of first set of three). Now we look at the middle splitter, it will attempt to split it's input 50-50 onto output, so half a belt worth at the third splitter, and half a belt worth at siding, and it succeeds at both. Notice what's happened: Despite the fact that we are supplying 2 belts worth of throughput to the balancer, it ends up finding an equilibrium that only assigns half a belt to the first siding and a full belt to the second.

Okay, I kind of understand whats going on with the splitters now. However I still think there's an issue with the inserters. Obviously the iron plate belts are backed up in the picture but I can tell you that the iron line doesn't have such big gaps when it's flowing down my bus as when the copper is flowing, and I don't understand why. Also the number of smelters are identical when I took that picture, there were 2 banks of 20 smelters for both copper and iron, so 40/40. Yet somehow the iron was more compressed.

[Frightning]

The first is a common problem: Inserters don't always manage to compress the belt, it's basically a matter of the timing of earlier insertions of plates causing the gaps between plates to be narrower than a single plate but still>0. One way to fix this is to find the inserter(s) that aren't being able to insert and make the belt under the tile they would insert on faster than the rest of the belt (note that this doesn't work w/ Express belts because they are the fastest possible belts!).

The second is an example of how the splitter behavior can be counter to intuition. The reason the first siding is only getting half throughput (did you notice that that belt is exactly half saturated?) is because of the splitter further down is having to put a full belt's worth of throughput out the latter siding. What's happening is this: The first splitter in the second set of 3 splitters will take 50-50 from incoming plates to feed the left exit (which is only one not plugged up). Following that fact backwards to the first set of splitters, we see that the last of said splitters is thus having to supply 1 belt's worth of throughput, which it will take 50-50 from inputs. This is key because that means the first splitter of the first set is going to put half a saturated belt worth of throughput onto the bottom output (which feeds into the last splitter of first set of three). Now we look at the middle splitter, it will attempt to split it's input 50-50 onto output, so half a belt worth at the third splitter, and half a belt worth at siding, and it succeeds at both. Notice what's happened: Despite the fact that we are supplying 2 belts worth of throughput to the balancer, it ends up finding an equilibrium that only assigns half a belt to the first siding and a full belt to the second.

Okay, I kind of understand whats going on with the splitters now. However I still think there's an issue with the inserters. Obviously the iron plate belts are backed up in the picture but I can tell you that the iron line doesn't have such big gaps when it's flowing down my bus as when the copper is flowing, and I don't understand why. Also the number of smelters are identical when I took that picture, there were 2 banks of 20 smelters for both copper and iron, so 40/40. Yet somehow the iron was more compressed.

The reasoning why is implicit in my previous answer: it has to do w/ timing of your inserters that unload the smelters, if they are synchronized nicely, then you can theoretically get perfect compression, conversely, there is a worst possible case where every plate is placed just ever so slightly too close for another to fit in between. In practice. you can end up anywhere between those extremes and more often than not, it will quickly find an equilibrium (there are many in this situation, I suspect). A similar thing is at work w/ the splitters in the second picture.

Both of these I find very interesting mathematically, I've been wondering if there's any practical way to engineer such problems away for the splitter case (belt case is easy if belts aren't express belts).

[XKnight]

Maybe this will help you:



11 steel furnaces produces = 11*2/3.5 = 6.28 iron plates per sec, while yellow belt throughtput is 6.66 items per sec.
The result of this underproduction is a small gaps on the belt, but belt is still compressed.

Blueprint

Code: Select all

H4sIAAAAAAAA/72Y4W6rIBTHX2Xxsy5V0bbZ5UmWG0PldCOxaACXNcZ3v9iuW1l6Wzwkflon/A/w45y/J/L2qWlr1jxVdABphBGg6TBIdgAaGcWk7lplkh00
Joq7VtsJraTDJ03SND7SJB9jLhTUp8flGHspV1jlFivcYIVrrLDECglWWGCFOVaYoTMAKcQmDnY97AGxRG/cfVy30qi2qXbwzj5Eq+hQC1X3wlQg2a6Bigs9
/aV71miIL4MKGK/emeTVFMCWtqZG9T/jl6fniYeWA12NXpvE5hm2IrAliK15rMmgbe2uk9p70oZJk9TtYSckM626aafZtZzcyJq9aAyoydy1eJOsoYM5dnaB
D6FMz5ooPq93HkwU8Gi0UXppbPUIycHucxynR1KeV7GhXqM0+ksHO9mGPb0+jpXgNCu+08z+N8l+MotDLTioR8dZPRePDvQVacpkLr42tBdKm2rOAc90aba2
vw8dU9OOaPQnitvedP28WN2xOhGr9qo9VEJa/anoUNjshV6PkYnha5R58Y7fFIB055TONeyZNomQGpRNiltv2Yf5dGUk/EuLpm8r4Jo+jf5D7Max7jHbOEfW
BqBJ9r2SrIbf/cFzcck6L0Tr5RGt0IhKh0q+8qZSzqRSLE5li4aycaFk3lDITCj54lA2WCg2NRwoxBtKNhNKujiUNRqKayp56Q1lNQ/K8kxKNBPiMvE22nQe
kuVrp0AjcV2WeLtsPg/J8h5L0EhcjyXeHlvMQ7L8yzjHIiGuwxJvh13PQ7J8C4dH4vor8fbX7Ux7Xd5f0V2tf16kMx311NynDgTvhhchK3CyHCdLUTKcCrdF
HA8cfNxNp3dwwGenQOvk8dddp8yymQHSwADbQP0mUL8O1JeB+iJQTwL1eaA+C02gMH1g/gauHnj4QPaBVx+YeYGJH1h3gWUf6DqhtnnLNcdY1KdG49cnX2Hg
8N25OL3F+P2pN76j+bWPi2jq718UGBvrqXoByf8Bhc9tr10cAAA=