The sense of the provider chest

[ssilk]

Well, I want to discuss it. I mean, there are three chest types for the circuit-network: storage-, requester- and provider-chest. The logistic robots move the items between those chests. The requester chest has the highest priority, then the providers and then the storage (correct me, if I'm wrong). The robots always try to empty a requester chest.

So, at this point I can use the providers to put some stuff in, which is then moved either into the requesters or storages.

I see the following problems here: you cannot tell, in which storage an item should be inserted - maybe I should speak better about the target destination. The stuff is stored in the nearest available a chest, which might get a problem, if that storage chest is full. Because that chest works as an output for another production.

Well, you can handle that, by wiring the inserters for the provider chest, but that's stupid: assume we don't have enough logistic bots to handle all transports, then items are inserted into provider chests. To count them correctly, you need to wire all storage AND all provider chests into one big network. This is stupid, because you can half the work just by using only storage chests.
The consequence is for me: I use only storage and provider chests and didn't see any disadvantage in that. The layout looks mostly so: the item from an assembly is moved by a smart inserter into the storage chest. The chest and inserter are wired, the inserter programmed to move only 10 items into this chest or so. So when the amount is reached, the inserter stops, production stops and all is fine.

I can think only for one use, where a provider chests is really needed: if you want to bring stuff from far outside into the inner factory, no matter, if there are more storages in between. For example ore or unused rests of ammunition. But who does this really need?

So the questions for me are:
- am I alone with this problems or did I just not see how they should be used correctly?
- is it possible to change the providers, so that they for example can be "targeted" into a direction? (The bot takes up the item and moves into that direction -if not requested by a requester chest - and THEN he searches for a storage chest), or have some other use?
- and if not: shouldn't they just be dropped from the game, because they make things only more complicated as needed?

[n9103]

Personally, I'd rather skip the storage chests than the provider ones.
With Storage, you can't control what goes into them unless you never use provider chests. With Providers, only what you want put into them goes in.
But either way, it's smart to wire them up to smart inserters to limit stockpiled factory output.

I might find a use for Storage should I ever get into a super-long game where I'd want to (pointlessly) make a huge central stockpile for thousands upon thousands of items, but that would just be throwing good resources after bad. (There's no useful point to a stockpile beyond what you fit on a belt between factories, making a giant one is all the more wasteful.)

[ssilk]

Personally, I'd rather skip the storage chests than the provider ones.
With Storage, you can't control what goes into them unless you never use provider chests. With Providers, only what you want put into them goes in.

Pardon, not quite sure, if I understood that correctly. You mean, the storage chests cannot be controlled on what goes into them, but the providers can? Eh?

But either way, it's smart to wire them up to smart inserters to limit stockpiled factory output.

I mean that's can be seen quite nice here in that pic: https://forums.factorio.com/forum/vie ... 1205#p8434
He wires all with green wires and uses all three types of chests.

I might find a use for Storage should I ever get into a super-long game where I'd want to (pointlessly) make a huge central stockpile for thousands upon thousands of items, but that would just be throwing good resources after bad. (There's no useful point to a stockpile beyond what you fit on a belt between factories, making a giant one is all the more wasteful.)

Hm. I don't use the storage chests for stockpiling (they are all nearly empty, only two or three exceptions) - or about the same usage as you seem to do with provider chests. I just use them that the logistic bots can put things in and out. I mean, that only storage chests enable to put things in AND out. Provider can only be filled with inserters and requester only emptied with inserters.

Code: Select all

      Provider   --->   Storage   --->   Requester

Logistic bots try to fill requesters first from providers. If empty they try to fill them from storage chests. Unused items in provider chests are moved into a storage chest.

And you know: the biggest advantage of the storage chests is, that all storage chests add their inventory in an own view. You see all items in all storage chests, if you hover over any logistic chest...

[grobyc]

From my point of view logistics chest systems looks like this:

provider-chest – used to put items inside from e.g stone furnace by inserters that can be obtained by logistics bots
requester-chest – used to put ordered items by logistics bots from logistic chest (provider or storage) that can be move by inserters to e.g. assembling machine
storage-chest – used to obtain items by logistics bots from provider chests (to store excess) if there is no task from requester chests OR used to obtain items by logistics bots from it if requester chest have order

I use only provider chest where I store items from e.g. stone furnace by inserters (no wires to limit items go inside) and requester chest to put ordered items to e.g. assembling machine (set limit inside chest depending from my imagination).

Stone furnace => inserter => provider chest => logistic bot => requester chest => inserter => assembling machine

I don’t use storage chest because I have a lot of all kinds of items in providers chest and logistics bots move all items to storage chests like they want and in one storage chest u can have one stack of all kind of items. I want to make stockpile only for few items in storage chests not all kind. There should be filter inside storage chest.

I think n9103 want to share same point of view by saing:

Personally, I'd rather skip the storage chests than the provider ones.
With Storage, you can't control what goes into them unless you never use provider chests. With Providers, only what you want put into them goes in.

Bots always try to fill requester chest first. If u have ordered items in storage chest and provider chest, bots will move items from storage chest first to requester and if storage chest is empty they start move items from provider chest. After requester chest is fill and bots have no other orders from requester chest they start move items from provider chest to storage chest. In this system bots perform double work, move items from store to order, after order is done move items from factory to store again. Its very realistic idea like in our lives but items haven't expiration date.

Finally from my point of view logistic system should work like this:
If u want build warehouse for items bots move items from provider chest to storage chest.
In provider chest and storage chest you have items with u want to move (order) to requester chest. Bots should move items first from provider chest to requester chest and if providers chest are empty in ordered items move it from storage chest. In this way there is no double work for bots.

When it comes to items in logistic system with u want to see: requester chest aren't count in logistic system storage. If I see 70k of iron ore in my logistic storage system I knew they are in provider or in storage chests and there is some count of iron ore in requester chests with I order and all is OK, I knew I have materials to go on. If I don't see iron ore in logistic system probably there is no iron ore in requester chest too or is about to run out and my manufacturing is/will stopped.
I don't want/need see items from requester chest in logistic system from one reason: logistic system show free/available items to use. Items in requester chest aren't free because logistic robots can pick up them. They are but not ready to use by logistic system, only by building or belts in the area where case stands.

[applemachine]

I found myself not using storage chests either. I thought I could use them to prevent the uneven distribution of resources in some areas of the factory, but the problem is that you can't control how much goes into storage. I use the wire system to limit the number of products being made so that resources are not needlessly wasted making too much of a single product. Is there a way to hook storage chests to the wire system so you can decide how much of each kind of item should be in the system?

[ssilk]

@grobyc
I can underline everything you said.
- storage chest need some filter, like requester, but by default they are without limit. Then it would be possible to have storages only for stones for example.
- the priorities of the logistic network should be tuneable (I thought the items are first moved out of the providers!) - or maybe it should be possible to control the priority (a window, where you can set the priority for delivering requester chests)
- the logistic system should show the number of items for all three types of chests and the three combinations (6 channels or filters)

And a comment: I currently mean, that this transport of providers => storage was thought to bring items as close as possible to the production, so that you can quickly produce in time. But you're right: the work has to be done twice, it's not like milk or bread. But this is a cool idea, if there would be items, which behave like milk or bread...

@applemachine:
I thought you can control very detailed, how much of anything goes into a chest: As said, I use the output-inserter of an assembly to limit any production - this has some reasons, for example I can work with the equal operator and you can use of course more inserters to output the production in different directions... I normally use red wire to connect all (!) chests/inserters into one big network and eventually I use green to limit it for single chests/inserters-pairs (eventually need an extra pole for that, that this pair-network is not connected to any other net). With this combination you can say for example if you have two electric-circuit assemblies: produce 2000 electric circuits for red and 1000 each for green. They will produce exactly 1000 items for each chest, even if you take out something from one chest. Of course you can st the limit for green to 1500 each, the one chest might be filled up with 1500 and the other only with 500. The point is, that this behaves much smoother, because there is not just "run or stop", its more controlled, more just in time.


And finally: connecting everything into a big network is time consuming. It would be a relief, when it would be possible not to wire that all up. - I think to "channels", something like a wireless network which works as a interconnected cable. Channel "L" for example is the sum of all logistic chests. Channel P are the providers only... And so on. (Or channels L has 6 "sub channels", which are the different filters)

All in all, till now my assumptions are more or less confirmed: only 2 of the 3 logistic chest are useful.

[n9103]

Yes and no. Storage does have a use, but it's pretty much limited to being a limit lifter.
If, for example, you find that your furnaces are backing up, and have been using provider chests for your outputs, then you could drop a few storage chests and, (assuming you have more than double your minimum needed bots to handle the increased flow,) your factories would all go at full speed, instead of working just to fill a limited space.
Not sure what you'd then do with the overflow, but I'm sure some sort of project could be imagined for the excess.
The reverse (using storage for outputs) doesn't provide any uses for a provider chest. (Other than a full renovation of your production line to use providers instead of course )

This would also help in measuring peak production, if you desired. This last one is probably the best reason for you to use providers, ssilk :p
You could get your setup just the way you wanted for your measurements, using lower limit providers at whatever bottlenecks, and you could start the whole assembly just by dropping a few storage chests (and a butt load of bots I'm sure)

[ssilk]

Thanks to all, who answered. i read that all many times and played the whole night, trying to find the point, and I think I have now an idea of what's wrong with the logistic chests.
First of all some abbreviations: provider chest : P, requester chest : R, storage chest: S, logistic network: LN, circuit network: CN.

The forum-members above mentioned, that they use only Ps and Rs. Yes, this works well, until your production stocks. I use smart inserters to limit the production then (Using a CN with connected P and smart inserters).
But then I found, that S has also its advantages.
And disadvantages. I describe what I did: I had a cloud of assemblies everyone connected with a R and P. The output smart inserter watches, that the production doesn't grow too much. For that all P and all output inserters in my factory cloud where interconnected.

And then I need more stone. I found stone only on the other side of a lake, very far around, it takes 13 coal to drive to that place. I build some solar powered stuff, built a battery of guns to protect the bots (using Rs to request the ammunition), and and I put just before every mine a P, because it was too far away for the belts around the lake and over the lake it was very near. Then I drove back. And indeed I had more than enough stone. Much than more than enough. The requester chests for stone was full. And the Ps on the other side weren't emptied any more. So I built some S, so that the stones would be transported over the lake. Not a good idea! The hell breaks, all stuff from the P where put into the S. I need to build many S and all of them need to be connected to the circuit network. Lot of work. :/

And after that, I had the problem, that the items wheren't in the chest near the assembly. They where spliced into several S. no chance to find the correct place in one turn.

So my conclusion is, that the problems begin with the S.
And even if i kow, that i better should wait for V0.7, I make some suggestions or naming of showstoppers for the LN/logistic chests:
1. it doesn't make sense that I need to wire every S. If I forget one, the whole system of complicated controlling of how much of one item should be produced can break down. There is either a limitation needed of how far a logistic bot transports something, or - in the other case, that everything in the LN is global - something like an autoconnected network only for S (see down). I think both together is not wrong.
2. I need some control about where is stored what. This means, that we could have two types of S, one without filter and one with. The Filter-S are preferred to fill and only if nothing is left, the stuff goes into the unfiltered. It would be great if the first 10 seconds or so the chest doesn't work (enough time to set a filter. this would be also fine for smart inserters!).
3. Provider chests will be always used like this:
. assembly -> inserter -> provider chest
or
. belt -> inserter -> provider chest ... And other combinations...
I don't see any other usage, cause P can be only filled by an inserter. So you have always to program the smart inserter to limit production. This is a lot of repeating work and if you want to increase the amount you need to find every inserter which controls this item and increase the number. It would be much, much easier, if something like a mega beacon /control center does this programming? You can control an area and it works like connecting every provider or storage chest to one CN and program the smart inserters so that they don't produce more than configured.

At the end I would say: all three types of chests make sense. The example above shows clearly, that I can do jobs with them, which wheren't be possible with 2 only. But the handling needs some changes, especially the role of the storage chest should be clearly defined.

[n9103]

I'm curious why you added a storage chest instead of additional requesters?
Until/unless Storage does become filtered a la Requesters or Smart Inserters, they don't have a good/efficient use.

[ssilk]

I wanted to bring the stones to my side of the lake, because I didn't know, how long the little defense could withstand the creepers. Using a requester means, I need also an inserter and a provider chest to move the items from the requester into a provider chest. And in the end: that's, what storage chests should be used for: storing stuff.

[n9103]

In the end, yes, but we're not at the end yet :p