Factorio Forums

mrvn wrote:Note: The 25 belts thing was stating a fact. Not challenging your math.

You need to challenge my math so that you can understand it. You need 17 belts to the trains. Bonus points: 17 is a prime number, you can't split it up into a smaller set of multiples.

dragontamer5788 wrote:No. Because no amount of belts can handle 677 gears per second + 967 iron plates/second + 805 iron ore/second in this amount of space or regularity.

I've repeated this over and over and over again in this thread. The numbers do not help your case at all. Now divide by 40, how many belts do you get per stage?

As for the balancing: You're assuming an end result that has no explanation. You start with saying your belts are already unbalanced. Then you say unbalanced belts will unbalance loading. DOH. You assumed the world is broken, ergo it's broken.

You need a balancer because you have 17-belts leaving the assembly machines, 21-belts coming out of the mines, and 25-belts coming out of the smelters. This innately causes imbalance between the lanes (all three are relatively prime against each other!!). Productivity3 modules really complicates the ratios here, but are necessary for maximum efficiency. The belts are unbalanced because ratios are unbalanced due to PM3 modules. Each stage of processing grossly complicates balancing efforts.

These simple explanations are why I'm frustrated by your rhetoric. Its outrageously clear to me that you haven't experienced this in game yet. I'm being visibly aggressive against you and your rhetoric because I'm at this point 100% certain that you do not have the experience to back up your words or assertions.

dood wrote:

Jap2.0 wrote:Additionally, you have to consider the extra time it takes to set up mass production of bots, roboports, solar panels, and accumulators, as well as the time it takes to expand enough to build and protect your power supply.

I don't think you should bring up time if balancing and individually spaghetti-ing 25+ belts is in play.

For this simple gears sub-base, it'd be 21-belt balancer (to balance the ores), 25-belts (to balance the iron plates) and 17-belts (to balance the gears). 63 belts to manage, and we're only at gears. Let alone the rest of the base. EDIT: Either that, or you choose to run slightly inefficient designs and instead spend more resources on PM3 / Speed3 modules, and the belts needed to connect them.

EDIT: Here's an alternative option. Build 20% bigger than necessary and allow for ~20% inefficiency in the belts (Ex: Accept the fact that some belts will only have an output of 32-items/sec instead of the ideal 40/sec). I will say that "building bigger to allow for inefficiency" is the simplest solution I can think of however. In the case of the 17-belt output, this means running ~21 belts to the trains. You'll still need a balancer on output to balance loading the wagons, but "building bigger" will allow you to avoid balancers in the intermediate stages. This is the approach I take in the mid-game (when resource expenditure still matters severely. But simplicity is more important than perfect efficiency). The Madzuri circuit-based balancer works extremely well if you are building 20% bigger and are willing to accept a slower loading speed. This allows me to build non-perfect balancers and still get a good job done in the midgame.

I also don't see the great debate, because four stack inserters can fully compress a blue belt...
(viewtopic.php?f=202&t=56504)

dood wrote:

mrvn wrote:Again, why would the input belts be unbalanced?

Ore patches are not perfectly rectangular and run out you know.

mrvn wrote:

dood wrote:

Jap2.0 wrote:Additionally, you have to consider the extra time it takes to set up mass production of bots, roboports, solar panels, and accumulators, as well as the time it takes to expand enough to build and protect your power supply.

I don't think you should bring up time if balancing and individually spaghetti-ing 25+ belts is in play.

I'm not sure how you play but when I need a 25 belt iron gear wheel factory I first build a simple one lane version. Then I blueprint 23 more of them in parallel. Then connect outputs and inputs to the trains with. No belt crossing or spaghetti-ing involved.

It's a bit more involved than just running belts into trains.
Unless you are okay with the train never leaving if one belt ran out of ore, that is.

That's why you balance the belts at the ore patches. You probably don't have ore patches with 24 blue belts of output so balancing there is much simpler. Doing 4 belts for a mine, balancing that and then loading into 4 car wagons works fine for me.

I was talking about the smelter, geer factory, circuit boards, ... If you make sure you start with balanced trains of ore you can design all your factories to keep that balance.

Are you guys seriously having a debate whether or not bots are better than belts?

I see you are discussing balancers and one way to deal with imbalance is to have overcapacity on the belts. Ie if you have 16 belts in and only need 15 out you can just make it 16 belts out and not require a balancer. However not all production centers can be balanced. For example if you produce both pipes and mining drills at your blue science place you'll need a balancer. Same goes for ore on the in-side. This is never even a consideration a bot user have to make though.

As for space the big difference is the flexibility in how you can space them, not really how big the footprint is (although bots are much more compact and usually with more beacons per assembler). With bots you can place single assemblers and roboports randomly and still be efficient. Try doing a playthrough where you ban cliff explosives and landfill. When your blueprints don't work form the get-go you'll really notice how much longer it takes to place belts.

mrvn wrote:That's why you balance the belts at the ore patches. You probably don't have ore patches with 24 blue belts of output so balancing there is much simpler. Doing 4 belts for a mine, balancing that and then loading into 4 car wagons works fine for me.

I was talking about the smelter, geer factory, circuit boards, ... If you make sure you start with balanced trains of ore you can design all your factories to keep that balance.

The math literally never adds up with +productivity in the mix.

4-wagons offload as 14.8 belts (double-sided unload) or 7.4 blue belts (single-sided unload). Immediately we already have an issue if you're going at max speed of 3.7 blue belts per wagon, because the belts come off the train without max compression, which implies that they will be unbalanced to some degree. (Note that 12x stack inserters wagon->chest handle 332.4 items/sec, but 12x stack inserters wagon->belt only handles 148 items/sec due to the slowdown. So you're already 50% slower than bots per train stop)

There are simpler designs than going full tilt. Slow down to 6-belt output for example for 1x side 4x wagon unloader with circuit network assist, similar to this concept. This will give you 240 iron ore/second. Cool. All's well at this point, a bit slower than usual, way slower than bots... but they're compressed perfectly and balanced perfectly.

Now what? You put them into the smelter with +20% productivity. Bamn. Now you have 288 iron plates per second. That's 7.2 blue belts of output. How do you possibly manage to balance 7.2 belts without a balancer? Spoiler alert: you need a balancer. Either that, or you slow down your smelting array to 6-belts of compressed output through the magic of priority, sideloading or whatever, suffering a 17% slowdown.

dragontamer5788 wrote:Immediately we already have an issue if you're going at max speed of 3.7 blue belts per wagon
...
That's 7.2 blue belts of output. How do you possibly manage to balance 7.2 belts without a balancer?

The way I would do 3.7 turning into 7.2 is have the incoming train offload 6 balanced belts. Split them just before the assemblers/furnaces into 12 belts, and load the output wagon with 8. When all belts start out balanced you don't need any complicated balancer to balance the belts. You just need splitters to split and recombine adjacent lines so the items can get around.

(really the way I'd do it is to have 3 belts out, and 6 in, and allow slower train loading/offloading, because if I really cared about speed bots are faster anyway)

Of course with bots you don't need to do anything. Just place a provider and a requester and its done.

mrvn wrote:That's why you balance the belts at the ore patches. You probably don't have ore patches with 24 blue belts of output so balancing there is much simpler. Doing 4 belts for a mine, balancing that and then loading into 4 car wagons works fine for me.

I was talking about the smelter, geer factory, circuit boards, ... If you make sure you start with balanced trains of ore you can design all your factories to keep that balance.

This was the example. It's straight up ore into gears and the ore field apparently has 25 belts of output because it is huge and speed modded to hell and back.
There are no "balanced trains of ore" to begin with.

Hedning1390 wrote:

dragontamer5788 wrote:Immediately we already have an issue if you're going at max speed of 3.7 blue belts per wagon
...
That's 7.2 blue belts of output. How do you possibly manage to balance 7.2 belts without a balancer?

The way I would do 3.7 turning into 7.2 is have the incoming train offload 6 balanced belts. Split them just before the assemblers/furnaces into 12 belts, and load the output wagon with 8. When all belts start out balanced you don't need any complicated balancer to balance the belts. You just need splitters to split and recombine adjacent lines so the items can get around.

(really the way I'd do it is to have 3 belts out, and 6 in, and allow slower train loading/offloading, because if I really cared about speed bots are faster anyway)

Of course with bots you don't need to do anything. Just place a provider and a requester and its done.

Sorry for the confusion. But its 3.7 belts per wagon -> x4 wagons -> 14.8 theoretical belts to unload from a 4xWagon solution. Which I simplified to 6x Belts in the middle of my rant.

This will give you 240 iron ore/second.

Because the easiest solution to the "belt balancing" issue is to simply compress your overly large belts into a perfectly compressed, smaller bottleneck. It slows you down, but its a grossly simple solution. The 6-belts of ores then turns into 7.2 plates of output.

Demonstrating that "Just because you're balanced at one stage, does NOT imply that you're balanced on another stage". Especially with random +20% or +40% bonuses to productivity all over the place.

Overcompression is a way to go there. If your 6 balanced belts of iron ore give you 7.2 belts of iron plates and you have problems balancing that to 6 train cars then just slow down a bit. Output only 6 belts of iron plates. You get perfectly balanced output while your input side gets a slightly slower consumption. So you aren't consuming 6 full belts of iron ore, so what? If you really need 7.2 belts of iron plates then maybe rethink your ratios. Go with 8 car trains and 8 lanes for example and keep the furnace count below maximum to get your 7.2 lanes output.

I think it's always better to design some over production into the factories. If I need 7.2 lanes of plates then I make 8. The extra 0.8 gives some extra buffer to handle spikes or congestion on the train network. Or if the iron ore mine runs dry and I setup a new mine the smelter can catch up and later stages in the chain might not even notice the gap in iron ore I had.

dragontamer5788 wrote:

mrvn wrote:Note: The 25 belts thing was stating a fact. Not challenging your math.

You need to challenge my math so that you can understand it. You need 17 belts to the trains. Bonus points: 17 is a prime number, you can't split it up into a smaller set of multiples.

dragontamer5788 wrote:No. Because no amount of belts can handle 677 gears per second + 967 iron plates/second + 805 iron ore/second in this amount of space or regularity.

I've repeated this over and over and over again in this thread. The numbers do not help your case at all. Now divide by 40, how many belts do you get per stage?

As for the balancing: You're assuming an end result that has no explanation. You start with saying your belts are already unbalanced. Then you say unbalanced belts will unbalance loading. DOH. You assumed the world is broken, ergo it's broken.

You need a balancer because you have 17-belts leaving the assembly machines, 21-belts coming out of the mines, and 25-belts coming out of the smelters. This innately causes imbalance between the lanes (all three are relatively prime against each other!!). Productivity3 modules really complicates the ratios here, but are necessary for maximum efficiency. The belts are unbalanced because ratios are unbalanced due to PM3 modules. Each stage of processing grossly complicates balancing efforts.

These simple explanations are why I'm frustrated by your rhetoric. Its outrageously clear to me that you haven't experienced this in game yet. I'm being visibly aggressive against you and your rhetoric because I'm at this point 100% certain that you do not have the experience to back up your words or assertions.

dood wrote:

Jap2.0 wrote:Additionally, you have to consider the extra time it takes to set up mass production of bots, roboports, solar panels, and accumulators, as well as the time it takes to expand enough to build and protect your power supply.

I don't think you should bring up time if balancing and individually spaghetti-ing 25+ belts is in play.

For this simple gears sub-base, it'd be 21-belt balancer (to balance the ores), 25-belts (to balance the iron plates) and 17-belts (to balance the gears). 63 belts to manage, and we're only at gears. Let alone the rest of the base. EDIT: Either that, or you choose to run slightly inefficient designs and instead spend more resources on PM3 / Speed3 modules, and the belts needed to connect them.

EDIT: Here's an alternative option. Build 20% bigger than necessary and allow for ~20% inefficiency in the belts (Ex: Accept the fact that some belts will only have an output of 32-items/sec instead of the ideal 40/sec). I will say that "building bigger to allow for inefficiency" is the simplest solution I can think of however. In the case of the 17-belt output, this means running ~21 belts to the trains. You'll still need a balancer on output to balance loading the wagons, but "building bigger" will allow you to avoid balancers in the intermediate stages. This is the approach I take in the mid-game (when resource expenditure still matters severely. But simplicity is more important than perfect efficiency). The Madzuri circuit-based balancer works extremely well if you are building 20% bigger and are willing to accept a slower loading speed. This allows me to build non-perfect balancers and still get a good job done in the midgame.

Right idea, wrong ratio.

With 17 output belts and 6 cars per train you would build it to produce 18 output belts. That way you get a nice 1:3 ratio. And yes, you would slightly under use the belts. BUT you wouldn't be inefficient on PM3 / Speed3 modules assuming the ratio of furnaces to train cars is nice. That's basically your design criterium. Pick a integer ratio of furnaces to cars. Then add the belts in a nice ratio even if they run below capacity. Belts are cheap compared to everything else.

I took up your challenge of producing a belt based setup. In this case I build the iron smelter with Bobs + Angels + Deadlocks loaders and stackers + Bulk rail loaders + LTN (everything else in the save game you can probably ignore). Look at the beautiful use of loaders for the initial packing of iron ore and the final splitting of iron plates. Imagine doing that with inserters?

Since Bob has larger beacons and the more advanced furnaces have more module slots I had to overshoot your target somewhat. 6 smelter already produce 58 blue belts of output and I build two sets for a total of 116 blue belts of iron plates.
Each furnace should produce 495.5 iron plates per second but that would be 95 recipe cycles per second and the game engine only allows one output per tick. So I kind of over beaconed it. Anyway, it's clocking in at 387 iron plates per second per furnace both with bots and belts. If you remove some of the excess beacons the belt setup could be made nicer and with loaders instead of inserters. It's kind of messy all crammed in the tiny area as it is.
Ore production is to the left using infinity chests. Ores and then stacked, merged into belts and dumped into bulk loaders. This part is way to big for the ore consumed but you should always have more mines than you need instead of need more than you mine. Normally this would be spread out over many mines so each mine would be far smaller. Probably just 4-6 lanes coming out of the miners, getting stacked, merged into one lane, split 6 ways and loaded onto the train.

One the right the plate production is unloaded, and split into 116 blue belts of iron plates. I'm producing a bit more (~8 extra plates per second) so the lanes are always compressed and the furnaces ocasionally have to pause for a cycle. Normally you wouldn't split it all into blue belts but dump it into the next factory directly.

Setting game speed to 100 and just looking at the FPS/UPS output on my screen the belt version runs at 190 UPS, the bots mk1 version runs at 100 UPS (so a clear looser) and the bots mk4 version runs at 200 UPS. The values fluctuate a lot depending on weather trains are loading/unloading so +-10 UPS is within the margin of error. Can't see a clear winner there for belts or bots in the game speed.

The whole smelter should run balance. None the less while figuring out how many bots I need to run full speed it became inblanced. So to get back on track and just in case I've added a balancer to the train stations:
It computes the average content of each bulk rail unloader and shifts items from one to the next in a loop whenever one has more than the average. Could have used the lowest tier of loaders and belts there as the throughput is way less than 1 per second.

For the bots version I needed 10000 logistic bots mk1 or 1800 fusion powered logistic bots mk4. You said resources are of no concern. Well, here are the cost for producing 10000 bots mk1 or 1800 bots mk4:
Now I see that I didn't break down copper wires, wooden boards, fibreglass boards, steel plates, ... fully. For bots mk1 the highlights re 50k steel plates, 120k copper plates, 140k iron plates. 28k plastic and 26k raw wood, .... That seems manageable but with mk1 bots the game crawls.

On the other side of the spectrum we have the bots mk4. While you need far less the cost are just staggering. 22M copper plates, 10M tin plates, 8M iron plates, 7.5M aluminium plates, 6M plastic, 5.5M raw wood, 3.7M lead plates, 3.7M titanium plates, 2M gold plates, 2M silicon plates, 2M coal, 1.5M iron ore, .... WOW Don't tell me those kinds of resource numbers can just be ignored. It also needs 4.1GW of energy (or 7.9k steam engines), 761 assembler mk6, 9.3k electronic assembler mk3, and some other stuff to produce that many bots over the space of 1 hour. All of that for just one train station. You need tons of them if you do the ore crushing, refining, sorting, smelting and casting for each ore type with bots. Although you only need one bot factory.

Savegames attached if you want to run it.