Train offloading

[Mr. Tact]

I built this four cargo wagon offloader below expecting it would result in even unloading of the cargo wagons regardlesss of the drain on each of the for resulting supply lines. The cargo wagons actually unload into chests, and this setup pulls from those chests.

However, that does not appear to be the case, one of the four subsets seems to consistently empty faster than the others. I can't figure out why. Am I missing something? Maybe you can't fully compensate for an uneven pull on the supply lines?

train offload.jpg (448.16 KiB) Viewed 11376 times

[ssilk]

Moved from General to Gameplay Help.

[Mehve]

A little tricky to follow everything there, you're definitely using more bits than strictly necessary. You've got two unnecessary underground belts in the first 4-to-4 balancer section (compare against my picture below), which are bypassing the pair of output splitters, possibly keeping the balancer from fully doing its job. Also, I don't see any particular benefit to any of the setup beyond that. You're doing multiple instances of splitting and then re-combining to balance the lanes out, but only the first instance is even necessary.

The picture below is a standard design that I use for unloading all my train cars now. The first part outputs one fully compressed blue belt of material from each train car, and does so evenly from each of the four chests, and has each chest supplying both lanes, which allows it to handle unbalanced consumption athe destination where necessary. Once you have a single lane from each train car, feed them all into the appropriate balancer and carry on from there.

Standard Train Car Unloader

[Mr. Tact]

My extra underground belts are simply setting up the lane balancer which is directly behind the load balancer. I wouldn't think that should make any difference. My load balancer is identical to yours. I should have thought to upgrade the setup with Stack Inserters though.. will consider that. But generally you have not spoken to the issue. Even if my setup isn't ideal, it should at least be "equal". But the chests for one of the cargo wagons "falls behind" in comparison to the other subsets. ie... more material is being pulled from the set of chests unloading from one of the wagons in comparison to the amount pulled from the others.

[impetus maximus]

you have balancers after the chests, then two more balancers. you only need one.
i use a setup very similar to Mr. Tact. four chests per car.

[Mehve]

He needs two levels of balancers. The first set ensures that the individual chests at each car unload evenly, while the second set ensures each car unloads evenly. At least, that's what's supposed to happen, the OP's problem not withstanding.

Just to clarify a few things, the train cars are all arriving fully loaded, right? No imbalance on the loading end of the train's run? If you block off the output until the train unloading chests are all full, and then run it, the chest contents definitely drop at different rates?

The reason I brought up those underground belt sections is that without all four lanes going into the last two splitters, if one of the outputs back up into the balancer far enough, it's possible to get imbalances, since each lane no longer sees the same number of splitters. It's a bit of a niche occurrence, but without seeing the device in action, I can only guess.

[Mr. Tact]

Just to clarify a few things, the train cars are all arriving fully loaded, right? No imbalance on the loading end of the train's run? If you block off the output until the train unloading chests are all full, and then run it, the chest contents definitely drop at different rates?

The reason I brought up those underground belt sections is that without all four lanes going into the last two splitters, if one of the outputs back up into the balancer far enough, it's possible to get imbalances, since each lane no longer sees the same number of splitters. It's a bit of a niche occurrence, but without seeing the device in action, I can only guess.

Yes, I have verified the trains are arriving full.

Correct. I have purposely halted the output of the load balancer and waited until all the chests were full, and a full train was sitting there waiting (four trains actually). And material is clearly being pulled from one subset more quickly. And I'm not talking about just a couple of ore.. after 15 minutes or so sometimes 3 of the cargo wagons will be empty but the fourth with be 90% full. It is pretty frustrating I have to say. I understand why load is pulled off unevenly due to the way inserters are loading assembly machiens, but I thought the balancer between the supply and the output would resolve the issue.

Actually, I've gone farther. I have basically allowed the factory to shutdown. All assembly machines not running because their output queues are full. All the furnaces have stopped, because their output queues are full. All the ore lines to the trains has stopped because they are full. And from this position, I still see the chests which are loaded from the cargo wagons get emptied unevenly.

In fact, it is doubly odd to me since the material is load balanced AGAIN, on the other side of the furnaces.

[Hannu]

There was a discussion some time ago. It seems that splitters work by such a way that they do not load inputs equally in different output loading situations and there is not simple and problem free general solution. I use usually circuit conditions to unload station storage chests evenly. It works but is quite slow even I use stack inserters (less than 1/2 of flow without balancing). However, one full blue belt from 6 chests should be OK with full stack bonuses.

[Mr. Tact]

Well that is a bit disappointing. *sigh* So how do people deal with this in large "mega-factories"? Seems like it could be quite an issue...

[orzelek]

It depends if you want to use mods or not.
Warehousing is very helpful for loading/unloading cine you can use one 6x6 or two 3x3 warehouses and balancing needs are much lower then.

[Mehve]

Kind of strange. Sounds like a pretty serious imbalance somewhere on the consumption side. You might also consider: 1 Feed the 4-4 balancer directly into a second 4-4 balancer, or 2) Change the 4-4 balancer into a 4-8 balancer, and then combine non-adjacent lanes?

[Hannu]

I have slightly different variations of this solution in high throughput lines. This one have 12 lines and are able to throughput continuously at least 9 full lines (probably some more but it is maximum consumption of my factory).
viewtopic.php?f=5&t=31809#p201355

In my opinion people seems to put huge efforts to achieve cosmetic perfection. I think that typical player of this kind of game would be an engineer (or at least think like an engineer) but they seems to be more like mathematicians. Everything must be perfectly balanced etc. But it is technically not necessary. For example my solution have enough parallel lines to keep production fully feeded 100 % of time even there are some imbalances. In many other cases it takes 10 % effort to get 80 % performance than 100 %. Or 12 % to achieve 100 % by adding some more units to make things. It may be not mathematically perfect but in my opinion imbalances and unpredictable variations are much more interesting and natural things than mathematically perfectly balanced machine in which every part works at 100 % efficiency. Unfortunately such elements lack almost completely from vanilla game.

[Mr. Tact]

I'll certainly agree people do make changes trying to achieve theoretical or visual perfection. And at some point, the issue would be self correcting.... If the first cargo wagon consistently has 90% of it's load left when the three other are empty, eventually the chests used by the other three will empty and all the load will end up coming from the chests for the first cargo wagon. However, I suspect by that point I would be having issues with less than four full lanes of plates flowing. It is hard to say how it would play out -- I am overbuilt a bit on purpose. I'm using the "LateUpgrades" mod and at this point all my available upgrades are in the thousands of science packs. I could probably test it by doing a couple of consecutive upgrades and simply watching without "correcting" and see what happens...

[Hannu]

If the first cargo wagon consistently has 90% of it's load left when the three other are empty, eventually the chests used by the other three will empty and all the load will end up coming from the chests for the first cargo wagon. However, I suspect by that point I would be having issues with less than four full lanes of plates flowing.

There will be certainly problem in so extreme cases. But if you control unloading of chests by circuit network, such a situation probably never happens, if incoming trains are full. At least I have never faced it.

I have also newer unload station. I have not yet used it much under very high load but it seems to work well at moderate loads or short periods of heavy load. I do not have count perfect balancers for every lines from buffer chests. Instead I connect the lines from the chests straight to the bus lines with simple splitters. If you have 24 chests and 4 bus lines you can feed every line from 6 chests and distribute load to many wagons. The circuit condition allows inserters take ore from the chest only if there are more that average-3 units left.

http://imgur.com/jrXTGio

I could probably test it by doing a couple of consecutive upgrades and simply watching without "correcting" and see what happens...

You are right. It is good way to get information if you let base run an hour or more at full load and just see what happens. Some imbalances are balanced automatically but some needs construction works and it is hard to say which is which (at least test is the fastest and less tedious way to find it out).

[Mr. Tact]

I updated my setup based in part on this conversation. I removed my lane balancer and added a second load balancer in series. It seemingly had no effect. Before completing approx. 10,000 science packs of research the unloading chests on the left most portion of this setup had emptied. Here is a screenshot of the modified setup.

train offload 3.jpg (540.16 KiB) Viewed 11102 times

Again, I have to say this is pretty baffling to me considering the load is re-balanced on the other side of the furnaces.

[Hannu]

You do not have the circuit conditions. Connect all chests together with red wires. Connect that wire also to input of an arithmetic combinator. Set input as copper ore / -16 and output as copper ore. Then connect all output inserters together with red wires. Connect this network (it must not be connected to other red network) to output of the combinator. Next phase is to connect all chests to the output inserters with separate green wires. Set inserter conditions to copper ore > -3. I am not sure if it can give 4 full belts from 16 chests but if you put 24 chests it will work.

Now the combinator calculates average content of the chests. It is subtracted (therefore minus) from every chest's contents. Inserters work only if number of ores in the respective chest is greater than average -3. Now, if the balancer takes more stuff from certain chest, its content decreases below average and its inserters begin to wait others. You can improve balancing also by mixing the lines. Connect one line from the first and one line from the second wagon to the first splitter. One line from second wagon and one line from third wagon to the second splitter and so on. The last splitter is feeded from the first and the last wagons.

I have 2 combinators but they do exactly the same thing. I learned later than I can reduce number of combinators by taking negative average and using a little bit different condition in inserter.

PS. I recommend to always use filter inserters to unload trains. I guarantee that when the first iron ore train stops to the station when you are on the other side of your base, you will not regret those 80 green circuits you have used to filter stack inserters.

[Mr. Tact]

Ah, okay. I see what you are saying now. An interesting approach, essentially force the uneven load to take from the supply evenly. I'll give it a whirl.

[Mr. Tact]

This does work. It is kind of interesting. The pull is still uneven but when the leftmost cargo wagon runs out, eventually the set of chests associated with it fall behind the others and stops supplying ore. However because the flow from the others is at capacity it doesn't take too long for them to catch up in the draining of their chests -- resulting in the rest of the train being unloaded. Here is a screenshot just before the last wagon runs out of ore. Pretty damn cool...

train offload 5.jpg (251.37 KiB) Viewed 11079 times

[Mr. Tact]

I am tempted to say it seems to me that splitter logic must be seriously flawed for this to ever occur. However, I suspect there is something I am overlooking. If someone can express what it is, I would be very interested.

[Ace_W]

as a new player, and a partial engineer... and someone who always tries to follow K.I.S.S. (Keep It Simple, Stupid)

I'm looking at distance. I think you are unloading to much into the system. The chests fill up too quickly and jam, so the belts jam, but because the left most rail car has to move the farthest it empties first.

What we are possibly seeing is a buffer jam. Try unloading from both sides of the car instead of one. See if the chests are filling too fast.

Are you running the biggest stack size for your stack inserters?