Stations & other LTN based designs

Adds new train stops forming a highly configurable logistic network.

Moderator: Optera

Post Reply
Copy link
Cadde
Fast Inserter
Fast Inserter
Posts: 149
Joined: Tue Oct 02, 2018 5:44 pm
Contact:
Contact Cadde

Re: Stations & other LTN based designs

Post by Cadde »

Thanks for the feedback!
I don't want one station per cargo as that would take up way more space. The (old)design can handle 31 different cargoes (though i would expect it to handle 6 or so without completely bogging down) which would take up at least 10 spaces wide per platform whereas the old design takes up 14 spaces wide for all cargoes. Say you have 6 cargoes. Per caro stations would take up at least 60 tiles in width and with this design it takes up 14 tiles in width.
Also, having all cargoes in the same <warehouse> makes it easier to manage upstream.

I made another iteration once i realized (or rather remembered again) i could abuse inserters to pick signals.

Image

Mods: bobinserters_0.17.10, WideChests_2.2.2, LogisticTrainNetwork_1.11.6, OpteraLib_0.1.8
READ NOTES FURTHER DOWN BEFORE USE
Also, WideChests (A.K.A Merging chests) need to be capable of handing 3x61 size warehouses. That is, max width/height must be set greater than 60 and you must set max chest area to 183 or greater (i have mine set to 256 max width/height and whitelist set to "1xN 2xN 3xN Nx1 Nx2 Nx3" and max chest area to 768)


Some things have changed...

- Signal picker is now entirely based on the "inserter trick" reading the hand contents of an inserter to pick one signal out of many.

- Due to how inserters work, picked signal will flicker on and off so effectively the signal picker clocks at 30 Hz. This can be remedied at the expense of increased complexity but 30 Hz is good enough.

- Sync circuit now starts every time inserters are triggered and counts for 5 "picker pulses" (Effectively 10 ticks, it can be PROBABLY be reduced to 9 ticks with a non-pulsing picker... At the expense of increased complexity)

- Estimated throughput per platform is now 24*12*6 = 1728 items/second. (The theoretical maximum is 24*12*(60/9) = 1920 items/second)
Of course there's other limitations at play here, such as train acceleration and the "2 second inactivity" in the trains schedule. But i would guesstimate the station (for a single wagon train and stack size 100) has a throughput of around 518 items/s or 11.5 blue belts worth.

- There is now a loading indicator. ;)

- Configuration is no longer necessary. As long as all available items exist in the warehouse on the right, the picker circuit will trigger the rest.

NOTE:
The station isn't hooked up to the warehouses, this station will NOT provide until you do that.
Also, there is a chest balancer using logistics drones but you technically don't need that. What you do need to ensure though is there is at least some extra left over in the left side of the warehouse as that side will always take the most items. (provide threshold could ensure this, also a constant instream of items could ensure this)

If you decide to not use the provided balancer (which technically isn't part of the station) you must ensure that the left side (that is, the trains right side) gets more items than the right or the left side will eventually unbalance itself as that is the side that adds the last 12 items to the train.

The design hasn't been tested on all possible train compositions or item combinations.
The last test is a L1C3 train (loco, wagon, wagon, wagon) with this schedule:
Image

Also note, there's no rail signals in this BP. That's an oversight on my part and technically the loading indicator isn't needed for the station to work. I usually put my exit signal right after the station and my entry at the last platforms medium pole. Throughput is everything!

EDIT: Oh BTW, there is one unused combinator that is a clock that starts running when a train enters the station. You can use this if you wanna measure the time spent loading or you can remove it. It runs on "enc positions of any loco" and outputs C. Where it is i will let you figure out but it's on the right side next to the loco.
Top
mrvn
Smart Inserter
Smart Inserter
Posts: 5969
Joined: Mon Sep 05, 2016 9:10 am
Contact:
Contact mrvn

Re: Stations & other LTN based designs

Post by mrvn »

This looks totally mad. I'm not sure I would ever want to place one of those for every station with multiple goods.

I can't accept anything where I have to adjust 31 combinators after blueprinting to match the 31 goods to ship at the station. The problem there is that there simply is no good way to select just one signal from many with combinators. Eliminating that would reduce your build a lot too.

One way around that might be to use a filter inserter configured to remove items from the warehouse and drop them back into the warehouse (pickup == dropoff). Configure the filter inserter to "set filter", "stack size = 1" and "read had content" and feed it the items missing in the train. It will pick up one of the items missing and emit the corresponding signal. Multiply the missing items count by that signal and you have the missing count for just one of the missing items.

Then use an arithmetic combinator with the "each - N = each" that you used before to set stack sizes so the selected item doesn't get overloaded.

PS: You can also use 3 rows of 7 inserters per side per wagon for a total of 504 items per cycle.
Top
Cadde
Fast Inserter
Fast Inserter
Posts: 149
Joined: Tue Oct 02, 2018 5:44 pm
Contact:
Contact Cadde

Re: Stations & other LTN based designs

Post by Cadde »

mrvn wrote: Mon Nov 25, 2019 3:03 pmThis looks totally mad.
Well thanks i guess?
mrvn wrote: Mon Nov 25, 2019 3:03 pm I can't accept anything where I have to adjust 31 combinators after blueprinting to match the 31 goods to ship at the station. The problem there is that there simply is no good way to select just one signal from many with combinators. Eliminating that would reduce your build a lot too.
Already did the inserter signal picker trick, see my previous post. Thus one no longer needs to configure 31 combinators as if that was such a big deal to begin with. Lacking the inserter trick i would still be using this big array of combinators as you only set it up once per item. It's not like you are sitting there micromanaging it and isn't so different from selecting the recipe on an assembler.
mrvn wrote: Mon Nov 25, 2019 3:03 pm PS: You can also use 3 rows of 7 inserters per side per wagon for a total of 504 items per cycle.
Using three rows of inserters doesn't make it much faster.
Let me explain why, by comparing the round trip time of inserters, here's the setup and it's timings:
Image
The numbers are...

[Num ticks to drop] > [Num ticks to reset]

Every inserter takes one tick to get items into it's hand from a chest. (which ofc is different to belts)
Then it travels the first number to the target chest and on the N'th tick the target chest received the items.
Then it travels back to the source chest and the tick BEFORE it picks up items is the second number.
Thus, the second number is the inserters cycle time.

There's two sets of timings for every inserter source->target configuration and that is because the offset matters!
In the first two source->target configurations the inserter head doesn't have time to adjust to/from the offset position before reaching a chest. Thus, putting the offset in the center eliminates this head travel. This of course only matters if the inserter takes less time to rotate than the head takes to adjust.
In this particular case it doesn't matter because we are comparing inserters that take well enough time to rotate to where the offset doesn't matter.

But it matters for the signal picker inserter i am using as that is picking and dropping from the same chest. A travel time of exactly one tick so the head should NOT move.

Right, let's compare throughput now then:
Image
Note: There's two extra rows at the bottom. These depict the very special (and absolutely highest moving throughput) case where you pick up diagonally like this:
Image

So as you can see, my current setup with 24 inserters per wagon has a maximum throughput of 1728 items/s. If i were to add the extra 4 inserters which can fit on each platform for a total of 28 inserters per wagon i could increase throughput by 288 to a total of 2016 items/s. An increase of 16.6% items for adding 16.6% more inserters. A linear relationship.

If i however were to change it to three rows of inserters and as you say 42 inserters per wagon i could move 2160.06 items/s.
My first thought on that is, as i said before "doesn't make it much faster", is that you only increase throughput by 144.06 items/s (an increase of ~7%) at the expense of 14 extra inserters or in percentage, 50% more inserters...!

My second thought on that is... Using three rows of inserters would mean another row of arithmetic combinators as to not overload the train wagon. This would increase the width of the station from my current 20 to 26. Just for 7% extra throughput.
And before you go "which is why one cargo per station is better" or any other thing like it i will reiterate that i don't want to build wide. If i wanted to build wide i could just as well run vanilla factorio and build wide. It's not difficult to build wide, it is however difficult to build compact and efficient both in materials and throughput. Which is the reason i have LTN, merging chests and bob inserters as those three are absolutely necessary for things to not grow super wide fast.
And also, before you go "so let wagons overload then". Well yes i have considered it many times already and simply moving unwanted items out of the receiving station. And every time i consider this i am reminded that overloaded trains will always block the provider station for (in my setup) 120 seconds.
...
Why do i keep the "stop timeout" setting at default? Because again, i don't like it when trains go and do whatever they damn well please because something interrupted them for a few seconds. But i have considered it and i might just do that rather than focus on not overloading trains.

But, these kinds of solutions are FUN. It gives the game a higher purpose for me. If all i wanted to do was to launch the rocket i would have done so a hundred times already and uninstalled the game by now. I always find something to improve upon and that is what makes the game for me.

So, there's also a fourth reason i don't use 42 inserters per wagon. Consider this scenario:
Image
How in tarnation am I to get power to everything? Yup, it leaves me back at overloading or one cargo per station... For 7% increase in throughput. WOOO!

And as if that wasn't enough reason not to use 3 rows of inserters. Symmetry is kinda important.
Let's start with the first problem assuming i am boxing all inserters neatly into a 7 tall area like this:
Image
Double using chests might not be such a bad thing really, especially with merging chests. But what if there were requester chests driven by logistic robots?
The idea here is consistency and you don't get consistency by having one chest empty before all others. (However, logistics robots aren't consistent anyways)
And that can't be fixed because...
Image
... You can't have three rows of 7 inserters delivering into a 2x6 area without double using chests OR changing their pickup radius (pickup radius being important for the cycle time as shown before, having one inserter with a longer cycle time among many means you have to synchronize to the slowest inserter and you lose a helluva lot of throughput that way)

I really hope this explains why i didn't go three rows of inserters and why i didn't even bother adding the two extra inserters i can add in the free slots in my latest build. (namely, 8.333% increase in throughput isn't worth the added complexity in the circuit and the lack of clear visual separation.)
Throughput is important and so is not overloading wagons. But throughput isn't so important to as where one would work with uneven numbers just to squeeze out another 8% when what really slows down the loading is the trains waiting for 2 seconds of inactivity after loading completes.

Thanks for the input though!

P.S. In case anyone plays with bob inserters but not merging chests... Moving items from chest to chest in a chain is almost the highest possible throughput you can get. Nearly as good as trains with bobs inserters.
Place a straight line of chests, one tile apart, and put three inserters either side. You get a throughput of ~486 items/second. So in a space of 7 tiles wide you get the throughput of 10.8 blue belts.

Do the diagonal trick and you can theoretically beat trains, because trains doesn't have a constant throughput like chests do. Somewhere around 700 items/s.
Do a very special setup with the diagonal trick...
Image
... and you can get 1800 items/s through 5 lanes that are "12 wide". That's equivalent to 40 blue belts!
Top
Cadde
Fast Inserter
Fast Inserter
Posts: 149
Joined: Tue Oct 02, 2018 5:44 pm
Contact:
Contact Cadde

Re: Stations & other LTN based designs

Post by Cadde »

HOLY QUACK!

I have to give props to mrvn for tempting me to explain why adding three rows doesn't do much. It lead me to figure out the throughput of diagonal trick inserters. My mind wandered to the idea of using the diagonal trick to load a wagon and this is what i came up with:
Image

The math then shows the reality of the situation.
12 items * 12 inserters * (60/4) cycles per second = 12 * 12 * 15 = 2160 items/second.
So using just 6 inserters per side, i can MATCH the performance of 42 inserters. I can even add a 7th inserter per side now as long as i am using merging chests.
With the 7th inserter per side, the throughput becomes 2520 items/s.

Oh well, back to the drawing board for the 7th iteration of multi cargo, no overload provider station.
Thanks... Really!

EDIT: There's one downside though... The chests are now in the way of the train stop so i can't extend them past the train stop. This means moving a lot of my factory around but it's worth it.

EDIT #2: I have a decision to make here now... I could add another 5 > 8 row of inserters for another 1080 items/s throughput (total 3240 items/s) for a 16 wide station instead of a 10 wide one. This begs the question though, is throughput THAT important? At 2160 items/s per wagon the only vanilla item i know of that can make that much of a difference is the space science pack at 2000 items/stack and technically speaking i am not very likely to make 2000 space science packs per second now am i?
The other stack size would be 200 and a wagon full of those (8000) would take less than 4 seconds to fill. And less than 3 seconds if i added the extra row of inserters. I find it hard to believe i am going to run into such problems anytime soon. At that point i am probably happy with my entire factory's throughput as is.

Not to mention i would have to have somewhere around 35 blue belts to constantly feed such a beast with just ONE type of item. I am more likely to have such immense production split over several factories or a single item station for just that production.

Also, going from 10 tiles wide to 16, adding the extra row of inserters for an increase of 50% throughput, would increase station width by 60%.
Simply adding another 10 tile station next to it would be more space efficient.

Nah, i am going with the single row of inserters. ;)
Top
Cadde
Fast Inserter
Fast Inserter
Posts: 149
Joined: Tue Oct 02, 2018 5:44 pm
Contact:
Contact Cadde

Re: Stations & other LTN based designs

Post by Cadde »

LTN Provider V0.7 (bob inserters, merging chests)
Iteration 7

Image

The area highlighted in blue is the actual station whereas the area in red is just the input into the chests.
Blueprint does NOT include the area in red.
Merging chests in the blueprint are unmerged, this allows you to rotate the station before placement. You need to manually merge the chests after placing the station down.

PRO
  • Matches requested items exactly, doesn't overload items, doesn't require locked slots.
  • Extremely high throughput, inserts 2160 items per second into each wagon.
  • Compact design, only 10 tiles wide, all circuitry fits next to the platforms and the train stop area.
  • Doesn't require any configuration, will automatically provide whatever is put into the merged chests.
  • Modular design, easy to extend for longer trains

CON
  • Uses mods (bob's inserters and merging/wide chests) as well as very quirky tricks that may change in future factorio versions.
  • "Too" compact design, getting items into the two merged chests at a high enough throughput can be a challenge.
  • Item balancing. Ensuring both merged chests have all items and equal amounts of said items can be a challenge.
  • Merged chests are directly adjacent to the railway track, making it impossible to put railway signals where the chests extend to, further complicating the balance between getting items into the station and having signals to allow trains to enter the station quickly as it becomes available.
  • Circuit not optimized for performance, doesn't filter all signals sending junk signals to combinators that won't be using said signals. It's this or a larger footprint. You are not supposed to have 500 of these stations anyways...
  • "Messy" layout, due to "last second adjustments" the circuit layout isn't neatly organized. The upshot is you don't need to change anything therein.
  • Might not work well with trains having more than 12 wagons.
  • Only set up for trains in L1C1-12 configuration. That means, one locomotive, 1 to 12 cargo wagons. It can be reconfigured for more locomotives but you have to ensure you reconnect the circuit output with the platforms.
  • Doesn't recover from "catastrophic" events/changes. Once a request has been triggered, the items in the station MUST exist or your train might get stuck waiting for the last item to load into the first wagon.

POTENTIAL ISSUES
  • As mentioned in "CON", if for some reason the merged chests (especially the one on the right hand side of the train in its travel direction) doesn't contain MORE than the requested items as a train stops at the station there is a potential for the first wagon (the cargo wagon closest to the train stop) to get entirely full. If this happens while the merged chests doesn't have all the items, the train can get stuck waiting for the last single item to be loaded due to the design using the first wagon as the "finalizer" wagon. That is, loading the very last item into the wagon if there's one item missing.
    This can theoretically also happen if the request is to fill all wagons. This because wagons 2, 3, 4, etc will never load single items.

    This can be remedied (assuming you are fine with underflow, as in the train not being entirely filled by request) by setting "stop timeout" to a low value in mod settings so the train doesn't sit in the station for two minutes doing nothing. I would think 8000 / 2160 * 2 = 8 seconds stop timeout would be a good value. This allows a full wagon of stack size 200 items to load in this particular design. But doing so might lead to underflow in other stations that doesn't have such high throughputs. See my earlier posts in thread for how to make all your platforms load items REALLY REALLY fast.


NOTE: The design can easily be adapted/changed to have logistic requester chests for each wagon instead of merging chests. Simplifying putting items into the station.

What Has Changed Since The Last Revision?
  • Found the 3 > 4 tick picker configuration while picking and dropping from the same tile. This means i can finally use a single inserter as my signal picker and no longer need an intermediate chest.
    Image
  • Because i am using ultra fast 3 > 4 tick inserters to fill train wagons now, it is no longer a good idea to use a clock for synchronization. So i changed my approach drastically.
    Once a signal has been picked, the entire requested amount is sent ONCE to a loop with this logic:

    Code: Select all

    w = cargoWagonCount();
x = w * 144;  // 144 items moved per inserter cycle times number of active platforms.
while (anySignal(allSignals)) {
  y = 0
  foreach (sig in allSignals) {
    y += sig;
    sig -= x;
  }
  sendToInserters(allSignals, y);
  wait(4);  // Delay for 4 ticks, this is done by a number of arithmetic combinators with identity signals.
  allSignals = selectOnlyPositiveSignals(allSignals);
}
    where the signal "y" is the new "s" signal because i needed the "s" signal elsewhere due to space limitations. Basically... Space...

    It means that exactly when the inserters have completed a cycle (4 ticks) the loop will have sent another pulse of the requested items - (144 * wagons) and this cycle continues until all signals are less than 1.

    While this loop is running, the signal picker is disabled waiting for the loop to finish processing. This is controlled by the "z" signal.
    Once the loop finishes (no more signals exist in "allSignals") the "z" signal is "released" and the picker will wait another 2 ticks (to allow train cargo to update and subtract from requested) and the picker will pick another item that remains requested.
  • Added automated "platform offset" so copy-pasting a platform (with this combinator) will ensure the circuit is autoconfigured for each platform.
    Adding more than 12 platforms is untested and should be avoided.
  • There are lots of "dirty" signals but they don't interfere with the operation of the circuit. Due to space limitations i couldn't fit filtering in there.
Adding More Platforms

Copy (or blueprint) the last two platforms. (Before copy, remove the train tracks or you won't be able to paste properly. Or use blueprint and right click train tracks)
Image
And paste so the last platform is overlapping.
Image
Everything should be ready to use. You can repeat the paste operation as many times as you'd like.
Top
mrvn
Smart Inserter
Smart Inserter
Posts: 5969
Joined: Mon Sep 05, 2016 9:10 am
Contact:
Contact mrvn

Re: Stations & other LTN based designs

Post by mrvn »

Cadde wrote: Tue Nov 26, 2019 6:02 am HOLY QUACK!

I have to give props to mrvn for tempting me to explain why adding three rows doesn't do much. It lead me to figure out the throughput of diagonal trick inserters. My mind wandered to the idea of using the diagonal trick to load a wagon and this is what i came up with:
Image

The math then shows the reality of the situation.
12 items * 12 inserters * (60/4) cycles per second = 12 * 12 * 15 = 2160 items/second.
So using just 6 inserters per side, i can MATCH the performance of 42 inserters. I can even add a 7th inserter per side now as long as i am using merging chests.
With the 7th inserter per side, the throughput becomes 2520 items/s.

Oh well, back to the drawing board for the 7th iteration of multi cargo, no overload provider station.
Thanks... Really!
When I suggested 7 inserters instead of 6 I assumed some merging chests or warehouse so "double use" wouldn't be an issue. On the other hand with requester chests the fill rate it determined by the number of bots on the way to each chest. Double the requested amount and a chest gets twice the throughput. So double use can be compensated by doubling the request. Within reason. Double using chests, merged chests or warehouses (size 6*X) would leave space for power between them instead of between inserters.

But back to the new setup using the diagonal setup. Your picture shows an empty row between the chest and the inserters. You can put inserters there for free. Problem is the combinators. They would add another 4 to the width.

Do you really need them all individually controlled? You can always use one decider combinator set to "each >= N => each" to control a number of inserters. Where N is the amount moved in one cycle by all inserters combined. So by giving up exact control of one inserter you can add another 14 somewhat slower inserters. That would give you 15 inserters controlled by the arithmetic combinator and 13 individually controlled inserters. Give up two and you get 2x8 inserters controlled by the arithmetic combinators set to turn off in 2 stages and 12 individually controlled. Not sure what the optimum is there without doing all the math. But my guess is the addition of 14 somewhat slower inserters makes more then up for the loss of exact control of one of the fast inserters.

Overall I concur that this becomes theoretical. For one thing optimizing loading time has diminishing returns as the time it takes to move in the next train remains the same and exceeds the time to load already. The other, as you say, is the problem of providing enough goods at that rate. Still, it's fun to design.

It's too bad the LTN station doesn't output the requested amount ahead of time. One could have 2 levels of buffer chests. The first generally buffers all goods. Then when the requested amount is shown the second level could be loaded count perfect in a less efficient way and then dumped into the train when it arrived. But generally you can't know which train will arrive next. So that isn't an option for LTN.

PS: As for using reserved slots. At 2000+ items a second and a 200 items stack size that would require 10+ slots reserved PER GOOD. So a 2 goods station would require 20 reserved slots or 50% capacity. 4 goods would require leaving the train empty, or something with a 50 items stack size.
Top
Cadde
Fast Inserter
Fast Inserter
Posts: 149
Joined: Tue Oct 02, 2018 5:44 pm
Contact:
Contact Cadde

Re: Stations & other LTN based designs

Post by Cadde »

mrvn wrote: Thu Nov 28, 2019 11:26 am When I suggested 7 inserters instead of 6 I assumed some merging chests or warehouse so "double use" wouldn't be an issue. On the other hand with requester chests the fill rate it determined by the number of bots on the way to each chest. Double the requested amount and a chest gets twice the throughput. So double use can be compensated by doubling the request. Within reason. Double using chests, merged chests or warehouses (size 6*X) would leave space for power between them instead of between inserters.
Yeah but thanks to you prompting me to experiment some more, the 1x6 block of inserters in iteration 7 now does what a 3x7 block would do. Essentially, you can't beat a cycle time of 4 ticks no matter how you twist and turn it.
mrvn wrote: Thu Nov 28, 2019 11:26 amBut back to the new setup using the diagonal setup. Your picture shows an empty row between the chest and the inserters. You can put inserters there for free. Problem is the combinators. They would add another 4 to the width.
Yes, i already considered adding another row of inserters and explained what it would take to make that work. Doing so only grants you another ~400 items/s and is thusly not worth the trouble. And getting 2160 (per wagon) items into the station from upstream is already quite a challenge. Even though the actual throughput of the station obviously is less than that.
mrvn wrote: Thu Nov 28, 2019 11:26 amDo you really need them all individually controlled?
Yes, no, maybe?
Yes because having them individually controlled means the fastest possible loading as you can eliminate one inserter cycle in various situations.
No because it really doesn't make much of a difference now that there's 15 inserter cycles per second.

mrvn wrote: Thu Nov 28, 2019 11:26 amYou can always use one decider combinator set to "each >= N => each" to control a number of inserters.
I can also use separate stack size signals for each inserter and thus anything > 12 items would mean all inserters can operate and anything < 12 items means just one inserter will operate. So theoretically i can have two combinators operating as many inserters as i'd like (limited by the number of available virtual signals ofc) as their minimum load is the number of inserters i have and anything between that minimum and the maximum can be individually controlled by signals.
However, this complicates the control circuitry. Might take a smaller footprint overall but complicating matters in absurdum isn't my cup of tea. ;)

The station (latest one) is already quite compact though and if i managed to get rid of a row of combinators either side i would have to push the control side lengthwise instead to actually make use of this slimmer width.
mrvn wrote: Thu Nov 28, 2019 11:26 amPS: As for using reserved slots. At 2000+ items a second and a 200 items stack size that would require 10+ slots reserved PER GOOD. So a 2 goods station would require 20 reserved slots or 50% capacity. 4 goods would require leaving the train empty, or something with a 50 items stack size.
Not sure i understand what you are getting at here. Merging chests (without altering mod settings) still have 48 slots per chest.
And in the test save i have now, loading enough stacks in what amounts to 12 chests for full wagons is 48*12/40 = 14.4 different cargoes. But there's more than 12 chests per side. And technically speaking i can extend those merged chests further to allow storage of more.
The downside with the station is you can't put the rail signal where you really want it.

I can probably use logistic chests instead and very close by buffer chests for the most frequent pulled items and in fact that is what i considered doing next. As i said, it would simplify upstream throughput. And allow me to place rail signals at every platform.
Top
mrvn
Smart Inserter
Smart Inserter
Posts: 5969
Joined: Mon Sep 05, 2016 9:10 am
Contact:
Contact mrvn

Re: Stations & other LTN based designs

Post by mrvn »

Cadde wrote: Thu Nov 28, 2019 12:24 pm
mrvn wrote: Thu Nov 28, 2019 11:26 amPS: As for using reserved slots. At 2000+ items a second and a 200 items stack size that would require 10+ slots reserved PER GOOD. So a 2 goods station would require 20 reserved slots or 50% capacity. 4 goods would require leaving the train empty, or something with a 50 items stack size.
Not sure i understand what you are getting at here. Merging chests (without altering mod settings) still have 48 slots per chest.
And in the test save i have now, loading enough stacks in what amounts to 12 chests for full wagons is 48*12/40 = 14.4 different cargoes. But there's more than 12 chests per side. And technically speaking i can extend those merged chests further to allow storage of more.
The downside with the station is you can't put the rail signal where you really want it.

I can probably use logistic chests instead and very close by buffer chests for the most frequent pulled items and in fact that is what i considered doing next. As i said, it would simplify upstream throughput. And allow me to place rail signals at every platform.
I'm agreeing with you that the reserved slots signal of LTN is not an option. It's slots in the wagons, not chests. Using the reserved slots at these speeds would mean the train has to drive around mostly empty in most cases just to handle the worst case.

If you are concerned with the size of your buffer chests you can add a second row of chests and inserters to double the size. The second row of inserter would do chest-2-chest transfere and could be controllef by the same signals as the first row. So that would only add 2 to the width.
Top
Ringkeeper
Fast Inserter
Fast Inserter
Posts: 143
Joined: Wed Feb 03, 2016 7:16 pm
Contact:
Contact Ringkeeper

Re: Stations & other LTN based designs

Post by Ringkeeper »

would like to have some opinion on depot design.

So far i build huge depots like on the first screenshot. But they need space and a lot of materials. Then i had the idea for a different approach.
Screenshot 2 shows my first design with 2 depot stations, screenshot 3 the current one with 4.
This way i can have more trains on the same space , as not always all are in depot (i guess on a running factory twice as many trains as stackers).

https://imgur.com/a/DIa5ClW?

Any idea to improve this? I need in total later around 500 trains (not in same depot :D ) and yes, i'm working with IDs, at the moment 2 networks (one for ore one for rest). Can't sell my wife more ID's quote "too much effort, don't want to change that always".....
Top
User avatar
Optera
Smart Inserter
Smart Inserter
Posts: 2920
Joined: Sat Jun 11, 2016 6:41 am
Contact:
Contact Optera

Re: Stations & other LTN based designs

Post by Optera »

Ringkeeper wrote: Wed Dec 18, 2019 9:52 pm would like to have some opinion on depot design.

So far i build huge depots like on the first screenshot. But they need space and a lot of materials. Then i had the idea for a different approach.
Screenshot 2 shows my first design with 2 depot stations, screenshot 3 the current one with 4.
This way i can have more trains on the same space , as not always all are in depot (i guess on a running factory twice as many trains as stackers).

https://imgur.com/a/DIa5ClW?

Any idea to improve this? I need in total later around 500 trains (not in same depot :D ) and yes, i'm working with IDs, at the moment 2 networks (one for ore one for rest). Can't sell my wife more ID's quote "too much effort, don't want to change that always".....
Unless your round trip time is over several minutes all your stackers are too big and too far from the depots to be efficient.

The best design is the final pic, but there's quite some room for improvement here too:
It could have decent throughput if the exit tracks are changed to allow 4 trains going on the main line simultaneously.
In and out tracks shouldn't share one intersection so trains going out don't have to stop for returning trains

I prefer 8 -10 track wide and 2-3 train long inline stackers so there's always at least 4 trains ready to go at any time.
The footprint of that is hardly larger than your design.
I know this is wider than 10 tracks.
I know this is wider than 10 tracks.
2018-11-02-08-41-10-5372923.png (326.69 KiB) Viewed 12780 times
Top
Hottemax
Burner Inserter
Burner Inserter
Posts: 13
Joined: Tue Apr 25, 2017 10:27 am
Contact:
Contact Hottemax

Re: Provider/Requester in one stop

Post by Hottemax »

It seems to me that both combined provider/requester blueprints from this post have a small error - the wires going from the stop and the output lamp to the arithmetic combinator that does the * (-1) are inverted. Input and output wiresshould be switched on the combinator, in order to match the behavior from the single purpose [prov/req only] blueprints.

After switching the wires the blueprints work great, thanks!


Optera wrote: Thu Jul 20, 2017 7:03 am
Basic Provider & Requester
This stop can supply/request any number of items.
To prevent timing issues (stop output is some ticks faster than the yellow combinator) it's waiting until train decoding signals are available before sending unloading signals.
Unloading logic uses 2 times unload amount to prevent unloading provided material once a train is slightly overloaded.


2019-09-25-09-09-32-8272336.png

Pro:
  • very versatile
  • fast loading/unloading
Con:
  • sorting multiple items from steel chests (I'm using merging chests and loaders)
  • inserters can lock up
    requires some tuning between stack size bonus and locked slots/wagon to prevent stuck inserters.

Logistic Provider & Requester for LTN 1.10.13
This is the version I use for my supply trains. Same principle as above, but it also makes use of train composition to activate only when a cargo wagon is parked in a loading bay.


2019-07-30-22-34-06-0663710.png

Pro:
  • extremely versatile
  • bots will automatically sort items from active provider
Con:
  • slow loading/unloading, only useful for small amounts
  • inserters can lock up
Top
User avatar
Optera
Smart Inserter
Smart Inserter
Posts: 2920
Joined: Sat Jun 11, 2016 6:41 am
Contact:
Contact Optera

Re: Provider/Requester in one stop

Post by Optera »

Hottemax wrote: Tue Feb 18, 2020 4:51 pm It seems to me that both combined provider/requester blueprints from this post have a small error - the wires going from the stop and the output lamp to the arithmetic combinator that does the * (-1) are inverted. Input and output wiresshould be switched on the combinator, in order to match the behavior from the single purpose [prov/req only] blueprints.

After switching the wires the blueprints work great, thanks!
Basic Provider & Requester was missing a connection from chests to the lamp. I'll upload a fixed version.

Logistic Provider & Requester works fine as is once a roboport is connected properly.
Top
conklech
Manual Inserter
Manual Inserter
Posts: 1
Joined: Tue Feb 25, 2020 6:13 pm
Contact:
Contact conklech

Re: Stations & other LTN based designs

Post by conklech »

Simple multi-item provider from logistics network

Image

This two-car provider station is designed to be easy to use for providing multiple items in small quantities out of a logistics network. Just configure any buffer chest to request a a single type of item from the logistics network. The combinators enable the corresponding inserter if and only if the item present in the chest should be loaded onto the train.

Here's how the circuit logic works. Each inserter receives two sets of inputs: the green wire indicates which items should be loaded (item signal = 1), and the red wire indicates which item is present in the corresponding chest (item signal = 1). The inserter is enabled if there is any signal equal to 2, i.e. if both inputs are set for any item.

Last edited by conklech on Tue Feb 25, 2020 6:25 pm, edited 1 time in total.
Top
burninghey
Fast Inserter
Fast Inserter
Posts: 133
Joined: Fri Sep 14, 2018 2:06 am
Contact:
Contact burninghey

Re: Provider/Requester in one stop

Post by burninghey »

Basic Provider & Requester was missing a connection from chests to the lamp. I'll upload a fixed version.

Logistic Provider & Requester works fine as is once a roboport is connected properly.
How to connect a roboport properly?
Top
User avatar
Optera
Smart Inserter
Smart Inserter
Posts: 2920
Joined: Sat Jun 11, 2016 6:41 am
Contact:
Contact Optera

Re: Provider/Requester in one stop

Post by Optera »

burninghey wrote: Thu Mar 12, 2020 1:06 am
Basic Provider & Requester was missing a connection from chests to the lamp. I'll upload a fixed version.

Logistic Provider & Requester works fine as is once a roboport is connected properly.
How to connect a roboport properly?
like this: viewtopic.php?p=297650#p297650
Top
khalismur
Long Handed Inserter
Long Handed Inserter
Posts: 53
Joined: Sun Feb 23, 2020 1:19 pm
Contact:
Contact khalismur

Re: Depot, Requester, Provider

Post by khalismur »

Optera wrote: Thu Jul 20, 2017 6:44 am
Depot for LTN 1.10.13
[...]



When importing this string, I get the following message:
Error while importing string:
Unknown virtual-signal name: ltn-position-heavy-locomotive
The import is apparently successful even though I get this message. Is it caused by another add-on I might be using or an issue with the BP string?
Top
User avatar
Optera
Smart Inserter
Smart Inserter
Posts: 2920
Joined: Sat Jun 11, 2016 6:41 am
Contact:
Contact Optera

Re: Depot, Requester, Provider

Post by Optera »

khalismur wrote: Sun Apr 19, 2020 7:45 pm
Optera wrote: Thu Jul 20, 2017 6:44 am
Depot for LTN 1.10.13
[...]



When importing this string, I get the following message:
Error while importing string:
Unknown virtual-signal name: ltn-position-heavy-locomotive
The import is apparently successful even though I get this message. Is it caused by another add-on I might be using or an issue with the BP string?
Sounds like I made that blueprint with signals set for Train Overhaul.
I'll update the string.

Edit: can't find any signal set to that signal. All signals are set to any-locomotive already.
Top
khalismur
Long Handed Inserter
Long Handed Inserter
Posts: 53
Joined: Sun Feb 23, 2020 1:19 pm
Contact:
Contact khalismur

Re: Stations & other LTN based designs

Post by khalismur »

Actually, there was an issue with the Blueprint. I think it was missing one wire from the Logistic Train Stop Output to the combinator that operates over the "encoded positions of every locomotive". This combinator wasn't getting the "encoded positions of every locomotive" signal and thus wasn't detecting locomotives to refuel.
Top
User avatar
Optera
Smart Inserter
Smart Inserter
Posts: 2920
Joined: Sat Jun 11, 2016 6:41 am
Contact:
Contact Optera

Re: Stations & other LTN based designs

Post by Optera »

Factorios bp import function could use some improvement.
When I took the blueprint a modded train was parked at the depot.
Factorios bp import function doesn't just skip missing signals, it doesn't place combinators containing them at all. However LTN does check for missing IO entities and placed the combinator itself, but had no idea of circuit connections.

I've updated the post with a bp and wrote an suggestion to import combinators regardless of missing signals.
viewtopic.php?f=6&t=83952
Top
khalismur
Long Handed Inserter
Long Handed Inserter
Posts: 53
Joined: Sun Feb 23, 2020 1:19 pm
Contact:
Contact khalismur

Re: Stations & other LTN based designs

Post by khalismur »

Thanks, I think your suggestion is a good idea. I've used a modified version of Odhrean's stations for years but decided to redesign my LTN+rails book starting off your base blueprints. Will report if any other small issue is found. Thanks once again for the continued support of your fantastic mod.
Top
Post Reply

Return to “Logistic Train Network”