Help me manage train unloading/understand logistics storage

[Yehn]

So this is the setup I'm using. When it's on its own logistics network, it works really well and I love it to death. Buuut...

If I hook it up to my main logistic network, it just gets weird. The active provider chests will send ore to storage far across my base (where there was none previously) instead of local storage. Assembler outputs will be carried to the storage field here, even when other storage is much closer. I really can't figure out the logic behind it, but more to the point...

Is there some way I can have my cake and eat it too? I can work multiple networks, but it would be more convenient if I could just have stuff on one network before I scale up to massive smelter blocks....

[siggboy]

I don't really understand the point behind this.

Do you want to create a very large buffer of resources first and then put it into a large smelting setup? Why?

[Yehn]

This isn't a very large buffer... it just keeps the trains moving. It's a non issue when the base is large enough to have dedicated smelting blocks with their own networks, I'm just looking for a better way to manage the in-between bit...

But that's the point: keep the trains moving, keep resources moving...

[siggboy]

Keep resources moving, but not consume them? I don't really understand what you're trying to accomplish by doing that.

To answer your question, though: it's not possible to directly influence where the logistic robots transport the goods. Maybe you should consider using requester chests instead of storage chests. Just have them request 20k ore each, and then use passive provider chests at the unload.

Later on, when you want to unload it, you can simply place passive provider chests (or storage chests) over the requester chests, and then reverse the loading.

[Yehn]

Well, so load varies for one thing. Depending on the sectors I have running my smelters might be going at it full tilt or not. The other thing is-- fresh outposts, the trains come often and quick. But then they slow down, or maybe run out before I notice... and setting up a new outpost takes a little time.

It is a buffer, basically to level variations in demand and supply. But there is always some at least some demand from the smelters...

That's a good suggestion for pure storage... but I'm already supplying smelters with the incoming ore so I can't implement it. I think I'll just have to split the network and try to bump up smelting on my todo...

[siggboy]

I don't know what your goals are with your Factory, but as a general principle you should not create large buffers of anything.

The problem with large buffers is twofold:

1. They're very difficult to move.
2. You can't see what the sustained performance of your production facility is. If the buffers are stocked up, then you're simulating the case where everything is in supply. But you do not know how much SUSTAINED supply you actually have -- i.e. how much your factory could produce with EMPTY buffers.

If you want to make X per minute (let's say 1 rocket every 5 minutes), then you have all sorts of resource demands to meet that goal. And you want to design a factory that can meet these demands while it's running.

It's not interesting to make a large scale production that only works if you spend 30 minutes beforehand filling up the buffers. Anybody can launch 1 rocket per minute if all the ingredients are stocked up to high heaven -- that's not very interesting (in my opinion).

[Yehn]

But... what do you do without a buffer zone? Just let the trains sit at the station, or let them go back to the mining outposts with non-empty cars...?

I don't really have any particular goals like rockets per minute. I guess right now, my goal is just to get high throughput through my rail network... (And then I'll probably beef up smelting, and then the main bus more if it needs it, and then power if needed, and so on...)

I've mostly got that goal, or at least high enough for right now (12000 ore per train, load and unload times at... 15 seconds ish?), but I was also hoping to work around the network split somehow... oh well!

[Frightning]

So this is the setup I'm using. When it's on its own logistics network, it works really well and I love it to death. Buuut...

If I hook it up to my main logistic network, it just gets weird. The active provider chests will send ore to storage far across my base (where there was none previously) instead of local storage. Assembler outputs will be carried to the storage field here, even when other storage is much closer. I really can't figure out the logic behind it, but more to the point...

Is there some way I can have my cake and eat it too? I can work multiple networks, but it would be more convenient if I could just have stuff on one network before I scale up to massive smelter blocks....

I don't know the full extent of your base setup, but I have experience with all logistic bot bases, and I have learned some things from it. You have said that you are using these storage chests as a buffer correct? So I have a question then: Do you have any Requestor chests in your main base's network? If so, bots prioritize filling Requestor chests before filling Storage chests, which would explain why they would move your ore all the way to your base rather than placing it in the nearby Storage chests. Furthermore, if the output side chests of your factory are Active provider chests, rather than Passive provider chests, this would explain why the bots are moving your other items all the way to your storage: Bots prioritize taking items from Active provider chests and moving them to Requestor or Storage chests (in that order) over other tasks.

[Yehn]

I don't know the full extent of your base setup, but I have experience with all logistic bot bases, and I have learned some things from it. You have said that you are using these storage chests as a buffer correct? So I have a question then: Do you have any Requestor chests in your main base's network? If so, bots prioritize filling Requestor chests before filling Storage chests, which would explain why they would move your ore all the way to your base rather than placing it in the nearby Storage chests. Furthermore, if the output side chests of your factory are Active provider chests, rather than Passive provider chests, this would explain why the bots are moving your other items all the way to your storage: Bots prioritize taking items from Active provider chests and moving them to Requestor or Storage chests (in that order) over other tasks.

Hi!

I do have requester chests-- for ore, requester chests > stack inserter > express belts feeding smelters (no modules or beacons yet). However, these are really close to the station.

I'm mostly using passive provider chests, but I do use active in a few places. For example, rail goes into an active chest since I have it set to stop making at a few thousand more than what can fit in a chest. I also put storage right next to my rail factory, but the logistics network just sticks it in the further away central storage instead. These are actual storage, i.e. yellow chests, not requesters.

And sometimes it would stick ore in my rail factory's storage, even though it was the furthest away of 4 storage locations and nearer ones weren't even close to full. (The rail storage area is very far from my smelters too)

All in all, the issue isn't that 'bots are moving stuff to all the way to storage', but that they prioritize very distant storage over near storage...

All these problems will go away late game, of course, but... *shrug!*

[British_Petroleum]

You could leave it as a seperate network, then use a power switch that you can use to power on/off a roboport which connects/disconnects this network to your main logstics network. When you're filling up the storage chests, leave the roboport off, and switch it on when you want to connect the supplies to the main network

[siggboy]

But... what do you do without a buffer zone? Just let the trains sit at the station, or let them go back to the mining outposts with non-empty cars...?

If the buffers at the unload are actually full, let the trains sit at the station until their wagons are empty. The great thing about the new train conditions in 0.13 is that you can now do this with Vanilla Factorio. You can also let them go back to the outposts with non-empty cars and let them do some monkey business driving around, but sitting at the station is fine if you have enough waiting bays so your railroad won't get congested.

Now, let's assume a "real" factory where you try to meet a production goal, e.g. 1 rocket per X minutes.

You need a certain amount of ore/coal/oil to meet your production goal -- and we're talking about sustaining the goal over an extended period of time, say, one hour or more.

The trains, then, have to continously supply the required amounts; the buffers -- at the outposts and at the main base where the resources are consumed -- are only there to bridge the latency gap that is caused by the trains requiring a certain amount of time to travel between outposts and your base.

In a perfect world (as I see it), all the buffers everywhere are hovering around at most 1 train load of material, and all the supply belts are sufficiently filled all the time.

You could even make sure that you keep your buffers capped at low amounts while your factory is not producing anything (i.e. rocket production is turned off), so that when you restart it you get precise readings of your sustained production without having to wait for your buffers to empty (which can take a long time with large buffers).

I guess right now, my goal is just to get high throughput through my rail network...

Sure, if you want to test the limits of your rail network, but you do not have any real demand, then you need to store the deliveries somewhere. That makes sense.

[Frightning]

I don't know the full extent of your base setup, but I have experience with all logistic bot bases, and I have learned some things from it. You have said that you are using these storage chests as a buffer correct? So I have a question then: Do you have any Requestor chests in your main base's network? If so, bots prioritize filling Requestor chests before filling Storage chests, which would explain why they would move your ore all the way to your base rather than placing it in the nearby Storage chests. Furthermore, if the output side chests of your factory are Active provider chests, rather than Passive provider chests, this would explain why the bots are moving your other items all the way to your storage: Bots prioritize taking items from Active provider chests and moving them to Requestor or Storage chests (in that order) over other tasks.

Hi!

I do have requester chests-- for ore, requester chests > stack inserter > express belts feeding smelters (no modules or beacons yet). However, these are really close to the station.

I'm mostly using passive provider chests, but I do use active in a few places. For example, rail goes into an active chest since I have it set to stop making at a few thousand more than what can fit in a chest. I also put storage right next to my rail factory, but the logistics network just sticks it in the further away central storage instead. These are actual storage, i.e. yellow chests, not requesters.

And sometimes it would stick ore in my rail factory's storage, even though it was the furthest away of 4 storage locations and nearer ones weren't even close to full. (The rail storage area is very far from my smelters too)

All in all, the issue isn't that 'bots are moving stuff to all the way to storage', but that they prioritize very distant storage over near storage...

All these problems will go away late game, of course, but... *shrug!*

OK, is there any ore in the far away storage? Because if there is, the bots will try to finish filling up those chests before 'opening a new one' (i.e. putting ore into an empty storage chest). They prefer to put items into chests that already have at least one of that item in it before putting it elsewhere.

[Yehn]

You could leave it as a seperate network, then use a power switch that you can use to power on/off a roboport which connects/disconnects this network to your main logstics network. When you're filling up the storage chests, leave the roboport off, and switch it on when you want to connect the supplies to the main network

Interesting idea, but I think the roboports' internal power storage would stop that from working precisely...

But... what do you do without a buffer zone? Just let the trains sit at the station, or let them go back to the mining outposts with non-empty cars...?

If the buffers at the unload are actually full, let the trains sit at the station until their wagons are empty. The great thing about the new train conditions in 0.13 is that you can now do this with Vanilla Factorio. You can also let them go back to the outposts with non-empty cars and let them do some monkey business driving around, but sitting at the station is fine if you have enough waiting bays so your railroad won't get congested.

Now, let's assume a "real" factory where you try to meet a production goal, e.g. 1 rocket per X minutes.

You need a certain amount of ore/coal/oil to meet your production goal -- and we're talking about sustaining the goal over an extended period of time, say, one hour or more.

The trains, then, have to continously supply the required amounts; the buffers -- at the outposts and at the main base where the resources are consumed -- are only there to bridge the latency gap that is caused by the trains requiring a certain amount of time to travel between outposts and your base.

In a perfect world (as I see it), all the buffers everywhere are hovering around at most 1 train load of material, and all the supply belts are sufficiently filled all the time.

You could even make sure that you keep your buffers capped at low amounts while your factory is not producing anything (i.e. rocket production is turned off), so that when you restart it you get precise readings of your sustained production without having to wait for your buffers to empty (which can take a long time with large buffers).

I guess right now, my goal is just to get high throughput through my rail network...

Sure, if you want to test the limits of your rail network, but you do not have any real demand, then you need to store the deliveries somewhere. That makes sense.

That does seem like a sensible way to approach it. Maybe, at least for now, I'll just have the trains sit. Although it is fun to have them zooming around...

I do think there is one advantage to at least some level of buffer, though: You can monitor it for trends. Like if your production is going at full speed and the buffer is getting leaner and leaner, that's a pretty good sign it's time to venture out and find more resources... before production chokes. But I do see the advantage of testing sustain from the start of the resource chain all the way to the end products too...

OK, is there any ore in the far away storage? Because if there is, the bots will try to finish filling up those chests before 'opening a new one' (i.e. putting ore into an empty storage chest). They prefer to put items into chests that already have at least one of that item in it before putting it elsewhere.

There wasn't, nope. When I was first setting it up, I even tried to prime the rail storage area by manually loading each chest with 1 rail, in the hopes that it would prefer that area for rail manufacturing output. But, nope, logi bots just went "F you I store things where I want!" xP

[Yehn]

If the buffers at the unload are actually full, let the trains sit at the station until their wagons are empty. -snip-

Oh, one last practical question. If I just have the trains unloading into passive provider chests, or maybe even have a second row of passive providers to give that '1 unload of buffer' thing... how do you keep the chests balanced?

Like for example... if the bots first empty the chests towards the front and then slowly work their way down the line, emptying just 1 or two at a time... I don't want the train stuck there when only the last few chests towards the end of the train have any ore left.

Do you normally just unload onto belts and balance that way? Or how do you prevent biasing..?

[siggboy]

Do you normally just unload onto belts and balance that way? Or how do you prevent biasing..?

I don't use robot based train stations a lot, so I'm not running into the issue that you've described.

The inserters will load the chests evenly, so the issue is with unloading from the chests into the logistic network. You can not influence the order that the bots take with that.

If your bots are slower with unloading than the trains bringing the resources, then you should probably unload into a row of normal chests (steel chests), and from there move the items into the provider chests for the robots.

The steel chests can be unloaded with a balancer: 23125

[Frightning]

If the buffers at the unload are actually full, let the trains sit at the station until their wagons are empty. -snip-

Oh, one last practical question. If I just have the trains unloading into passive provider chests, or maybe even have a second row of passive providers to give that '1 unload of buffer' thing... how do you keep the chests balanced?

Like for example... if the bots first empty the chests towards the front and then slowly work their way down the line, emptying just 1 or two at a time... I don't want the train stuck there when only the last few chests towards the end of the train have any ore left.

Do you normally just unload onto belts and balance that way? Or how do you prevent biasing..?

I haven't worried about biasing. The chests can each hold 48 stacks and the train cars only have 30, (I hear that's gone up to 40 in 0.13 but even so), a train car can't even fill one chest, let alone 12 or 14. So if you're unloading chests are getting plugged up, either your storage is full and have no requests for what's in them, or it's time to make more bots to increase throughput so they can actually keep up with the trains average unloading rate. The only times my unloading chests have been plugged up so far has been due to storage being full (happened quite a lot because I've been over-mining Stone and more recently Coal, but I've kinda not cared because I want a reason to fill in the gaps in my layout with storage; I have over 200k Coal and about 180k stone in storage, along with ~50k Copper which peaked near 100k hours earlier). The reason I wouldn't worry about biasing with the way I do bot unloading stations is that because I use active providers for the unloading chests, which are the highest priority job for the bots (they unload those before doing just about anything else)

[Yehn]

I haven't worried about biasing. The chests can each hold 48 stacks and the train cars only have 30, (I hear that's gone up to 40 in 0.13 but even so), a train car can't even fill one chest, let alone 12 or 14. So if you're unloading chests are getting plugged up, either your storage is full and have no requests for what's in them, or it's time to make more bots to increase throughput so they can actually keep up with the trains average unloading rate. The only times my unloading chests have been plugged up so far has been due to storage being full (happened quite a lot because I've been over-mining Stone and more recently Coal, but I've kinda not cared because I want a reason to fill in the gaps in my layout with storage; I have over 200k Coal and about 180k stone in storage, along with ~50k Copper which peaked near 100k hours earlier). The reason I wouldn't worry about biasing with the way I do bot unloading stations is that because I use active providers for the unloading chests, which are the highest priority job for the bots (they unload those before doing just about anything else)

Oh, yes, I wouldn't be worried about active, those get cleared really fast. Passives are a different story though...

[Frightning]

I haven't worried about biasing. The chests can each hold 48 stacks and the train cars only have 30, (I hear that's gone up to 40 in 0.13 but even so), a train car can't even fill one chest, let alone 12 or 14. So if you're unloading chests are getting plugged up, either your storage is full and have no requests for what's in them, or it's time to make more bots to increase throughput so they can actually keep up with the trains average unloading rate. The only times my unloading chests have been plugged up so far has been due to storage being full (happened quite a lot because I've been over-mining Stone and more recently Coal, but I've kinda not cared because I want a reason to fill in the gaps in my layout with storage; I have over 200k Coal and about 180k stone in storage, along with ~50k Copper which peaked near 100k hours earlier). The reason I wouldn't worry about biasing with the way I do bot unloading stations is that because I use active providers for the unloading chests, which are the highest priority job for the bots (they unload those before doing just about anything else)

Oh, yes, I wouldn't be worried about active, those get cleared really fast. Passives are a different story though...

Yep, that's exactly why it's recommended to have train cargo unloaded into active providers, so the bots will clear it first thing, thus ensuring smooth operation of your unloading station.

[MnHebi]

Personally, I would just remove that roboport line between the storage chests, and the providers. Use both inserters *and* bots to feed a requester chest line, and again have a line of inserters feeding a line of storage chests, and again after that line feeding another line. Only way you will have ore in those storage chests so the bots recognize them to be filled with ore instead of going for far off chests. The requester chests should be set with max request for ore, while the smelter setup should be setup with only 95% demand for ore. Yes, you will need more space for same amount of chests, but you will unload way faster, and way more efficiently than using bots alone, and trying to rely on bots being smart about their priorities. Cause they aren't, you need very smart choices about how much demand a requester chest has to make the bots smart about their tasks. If a far off chest has high demand, they will go fill that first even if it is on the other side of the planet.