I agree with the mindset of designing for guaranteed flow, but I completely disagree with the idea that 'avoiding balancers' is the way to achieve that. The balanced belt example at 2:50 gives 8 belts of variable output. Your method at 5:00 to manually cluster and balance etc, gives 5 belts of guaranteed flow. It's not the fact that you avoided the balancers that's important, it's the fact that you're only trying to extract 5 belts of output. If you had just used an "8 to 5" balancer to compress on the first setup you would get the same 5 belts of guaranteed flow. Less manual work trying to cluster things. Less worry about lane 1 having too many edge miners running out. No 'Spare' miners that aren't capable of outputting. Using the balancer is less work with more time of guaranteed flow because you'd only fall below 5 belts of output when the entire ore patch is incapable of providing the 5 belts of output. In your manual setup you could have belt 1 running at 95% throughput because it happened to have two edge miners that stopped earlier than the rest. Especially because in your example the 'spare' miners you don't use could have been looped into the balancer to provide input when others start failing.
Not strictly correct for the last part, you still can have less than maximum output if you are limited by the throughput of a single lane. Also, left and right lanes essentially work as two whole separate networks, so you can have one runout before the other. Other than that I agree.
For the first point you would need to make sure the balancer you use is throughput unlimited. The second point is correct but it is extremely unlikely that one side runs out significantly faster than the other. If one side is out, the other is often not far behind@@AbsoluteHuman
@@knicknacks That's just false, not all orepatches are symetrical, actually most of them simply aren't. On the outside the ore patches almost always will have less space for miners. With a belt balancer you can still use those sides without issue, without them you either need to create a complicated circuit network to somehow still balance the output. Or just merge those outer belts into one of the other belts.
@teaser6089 You could slap on a single belt lane balancer on the belts that are significantly sideloaded before they go into the balancer. Or on all the belts before the balancer, for that matter, if it's a big issue on that particular orepatch.
9:36 See this is where your entire argument for not using balancers because of "variability" falls apart. The reason you are able to achieve perfect ratios is because the belts have a maximum throughput, which is defined by the type of belt. It doesn't matter how you get a full belt of ore, all that matters is that you have a constant full belt of ore. The only difference is how the system behaves when it starts to run dry, with no belt balancing, you are looking at some parts becoming variable before other parts, since they don't all have the same number of miners. Whereas, with belt balancing, the ones that have excess miners will pick up the slack for the lanes that are mined out, at least until you hit the critical point that you simply don't have enough ore being produced to oversaturate the system. But at that point, the mine is being mined out, which is always going to happen, regardless of your methods for extraction. Essentially, your argument is that belt balancers increase variability, when in reality they reduce it. Also, 9:47 shows me why you don't like belt balancers, because you for some reason think that if you have 8 lanes being mined, you have to have 8 lanes being output. That will as you say induce variability very quickly. When I make a mine, I build as many lanes of miners as I can, then feed them into a N x 4 lane belt balancer. Which essentially gives the exact same result as what you are trying to achieve, except it's balancing across all the miners, to decrease flow variability in the system. On the topic of trains introducing "variability", that is simply a matter of train network design. If you assign the proper amount of buffering, as well as the proper number of trains for the distance traversed, you should never have a lack of resources at the unload station, unless the load station isn't producing enough, which is just a matter of output not meeting input demand, belts have the same issue.
Ok, I have two questions. 1. How many different belt balancers do you generally use? 2. How do you achieve a perfectly even flow when the number of inputs and/or outputs is odd? The train thing... I don't do buffering. The way I see it, if I have a train waiting for another train to be fully unloaded before it can do anything, why not get busy unloading it instead? Like instead of it waiting there buffering, construct a dedicated station for each and every train so that no train ever has to wait ever again. Wouldn't that be faster? That's the principle I use to design my rail networks. It is chaotic and it is highly variable.
@@theotherbigfoot So, I use prebuilt belt balancer blueprint books that I just slap into place, they have every combination of inputs / outputs you could possibly need. Using the blueprint books, if I have 7 lanes of ore being produced at a mine, I just slap down a 7x4 belt balancer blueprint. The people that made the blueprints did the work to allow every input lane to flow to every output lane. And that's the beauty of balancers, as long as you have 4 belts worth of input (for example 8 half full belts of input) then you can keep 4 output belts full. On the topic of trains, I was mostly referring to having the right amount of buffer chests. But your comment has me thinking again that yes, there are actually two ways of buffering trains. If you prefer to not have multiple trains per station, that's perfectly doable, but from a variability standpoint, you are less likely to create variability if you oversaturate the system. It's the same principle behind you routing the mining belts through two times, it creates oversaturation, ensuring the belt is full for longer. Having extra trains running per station is the same concept, just with trains. But you can also just make the train longer, which equates to more buffer chests unloading at once, which leads to longer times between the station being empty, and longer allowance for the train to fill back up and get back to unload.
@@theotherbigfoot I personally just use one 4-4BP, 8-4 BP, and a 12-8 BP, since my trains are 4 wagons long. As for buffering there's an extremely simple setup for train outputs: 6 blue inserters pulling out of the train car into 6 steel chests that have 6 more blue inserters putting them onto a belt. Eventually these become stack inserters if it's an extremely major station (for example the ore dropoff at your foundry). There are better setups for stations, but having this as your loading and dropoff is old reliable and does the job plenty well. Having multiple trains on a line can very rarely be useful, too, but you should almost NEVER have a line long enough where the train hasn't brought another load in by the time the buffer chests are empty. Not until the end end endgame
You seem to be overlooking a factor in that "designing for maximum flow" argument. (At least when ore trains come into play.) You are correct in that *the act of balancing* won't solve the variability of the mines, only *compression* (i.e. overbuilding miners) will. Which also means that if you belt directly from mine to factory, then you *can't* use an 8-to-8 or 4-to-4 balancer (to solve that problem). Because as soon as a single one of the input belts drops below a full belt of throughput, the output belts can never be full. The presence or absence of a balancer can't change that. In that situation you need a 12-to-8 or 10-to-8 or 8-to-4 balancer to guarantee that your belts are compressed (or more than 46 miners on _EVERY_ belt). But the trains make the difference: If you look at *one* outpost with an 8-to-8 balancer that loads into 8 wagons, then you can deduce that you'll almost never get the full 8 belt throughput going *into the train.* At first glance this _looks_ just as bad as directly belting ore *from miners into the factory* through an 8-to-8 balancer. But even if the train doesn't *load* at it's full 8 belts, it can still *unload* at a full 8 belts. That's where the depletion of ore patches comes in to save us. Since players know that their ore patches will deplete, they tend to future proof by connecting more ore patches than they need at the moment. But that also means that more trains are loading (at under 8 belts throughput) than are unloading. In total, this _usually_ increases the rate at which ore can flow *into trains* to above the rate at which the dropoff stations take ore *from the trains.* (And _if_ the input rate is lower than the consumption rate, we call that "I need to add more mines".) This overbuilding (more ore *patches* into the dropoff stations) provides the same compression as a 12-to-8 belt balancer in ore-patch-to-factory belts, or as if more than 46 miners are connected to one belt. The main difference is that if you overbuild ore patches, and if you use balancers that compress down, like 12-to-8 balancers, then you will have a balancer between the ore patch and the train, but not with the "connect 56 miners to one belt" strategy. And with any balancer there, even an ore patch where only a single belt still carries ore can contribute to that input flow rate. Otherwise the train won't be able to fill all of its wagons.
I don't disagree with anything you say. All I'm doing is setting out my argument for why I don't use them. My argument isn't necessarily right, it's just my argument. I would really like to see someone put together a video titled 'why I do use belt balancers' because I think it would help me understand the issue better. I struggled to come up with reasonable use cases for this video; I don't use belt balancers, I've never used belt balancers, so trying to come up with good examples was nigh on impossible. I wouldn't even mind if it meant I had to remake this video, if there is something I have missed then I'd be happy to update it. But what I'm reading in all the comments is that everyone seems to have a different take on the subject. I'm not reading one argument, I'm reading many different ones. Some even agree with me in not using belt balancers but disagree with why. It seems to be a genuinely polarising discussion. Which I love by the way. Its awesome that not only do we all have different approaches but that we are also willing to spend a lot of time making comments on youtube that explain those different approaches. The trains thing is something I want to cover in a dedicated trains video. It's relevant to this story, so I needed to mention it, but it isn't the subject I wanted to cover in detail. It isn't unreasonable to have expected me to cover it in more detail in this video, but hopefully once the trains video is done it will make a bit more sense. It might be a while though, I have at least 4 or 5 videos to do before then.
@@theotherbigfootin my big bases I have trains for pretty much everything, stations with multiple inputs sometimes multiple outputs, I can't do this directly to train (on the scale that i like), and i want to have more than 1 belt worth of inputs so I use balancers (for input and output, though output is much more important, unbalanced trains are a pain :D)
Using a balancer with a large compression ratio (8:4 for instance) makes setting up that "guaranteed flow rate" much easier to automate. Instead of carefully adding a few miners to edge rows, balancing out the number of miners per belt, etc - just overbuild by ~1.5x and compress down to a balanced flow with a balancer. The balancer ensures that the central miners pick up the slack for the edge miners as they exhaust, and keeps a consistent flow going. You get the same result as your guaranteed flow rate miner layout - with a proper balancer and enough input lanes, the output will remain 100% compressed until the or patch is physically incapable of producing enough ore to meet demand. Edit> Say you have an ore patch that can sustain ~150 miners. You could manually group them into ~3 50 miner groups with some careful arrangement so all of them will run out at about the same pace. Or, you could group them into ~5 30 miner groups and compress down with a 5:3 balancer and get the exact same result - 3 output lanes which have extra "redundancy" so they will maintain their consistent output for an extended period of time.
For me just always use buffers everywhere, i know its not really what this video is about, but by buffering you cancel out those times where your patch "overproduces" so you just have to worry about "is my buffer full" if yes, the factory works on 100%. This also kills the uncertainty about trains, if you unload a train into some chests, the chests will always unload at the given speed while my train network can be chaotic
Imho the only place where buffers make sense are in train stations and energy production. Elsewhere the belts and assembler contents are more than enough buffer for most purposes.
@roderik1990 if you use steam storage for nuclear, take a time to calculate heat capacity of all of the components that transfer that heat. I've got that 2x2 reactor with minimum (at least I really tried) h-piping and ceil ammount of heat exchangers in summary enough to store all of the heat that will be produced by one load of 4 fuel cells in each reactor and even slightly more than that. That fact elliminates the need for storing steam if you don't plan to use it as a battery for energy consumptions higher than reactor can produce.
@@roderik1990 yup you should do your manifolding at dedicated depots where resources are distributed. not at each individual input site where those resources are used. that's just wildly inefficient. you want your factories to have monoid like modularity.
Much like silos in a real factory. The incoming rate of raw material is variable, inconsistent. Building a reserve and pulling from it at predetermined rates allows downstream equipment to run efficiently.
This video is just... wrong. You don't use balancers to maximize output. Maybe at mining stations, but balancers are used at train unload stations because of the simple fact that it's virtually impossible to draw exactly evenly from the numerous belts coming out of your train. You are creating the inconsistency problem you are trying to avoid by refusing to use balancers.
Tbh, if you dont like balancers - use aai boxes and loaders. But its just my experience, and i dont find belts intriguing or interesting enough to mess with balancers
That said, the video isn't completely wrong, because you don't need balancers in mining outposts, flow routers (throughput-unlimited widgets) work just as well. Belt balancers are still quite useful after unloading trains though, but may not be necessary in rail city block designs.
What you are doing achieves what a balancer will achieve, only with more effort, thought, time, and variability. In your example ore patch, you are linking a specific set of miners on the outside of the patch to a specific smelting column. If you link that specific smelting column to a specific green circuit build, then, when those miners inevitably run dry before the center ones, your specific green circuit build will run dry. It's possible that the rest of the miners will be sitting idle, since the rest of your iron might depend on those green circuits to be flowing in order to be used. If you used a balancer, then all miners will be supporting all furnaces. When some run dry, it won't matter, especially if you apply your "gut feeling" system to the entire ore patch. Your ore patch has 292 miners available as a maximum. This is 4.2 blue belts or 6.3 red belts. I tend to stick with red belts for a much longer time than I should, and you're using them in this video, so I'll use those as my example. Feeding these 8 belts containing a total throughput of 6.32 red belts into an 8-to-6 balancer will give you 6 fully compressed red belts for at least a little while. This is what you're doing, only this abstracts the whole patch into those 6 belts rather than connecting specific rows of miners to specific belts. Even better, use an 8-to-4 balancer and have a guaranteed 4 red belts of iron for a long time. That's what I do. (Also, I included the 16 miners you very fairly labeled "annoying." You asked someone else how they'd handle an odd numbered belt. With this, I'd just lane balance it (your "scrambling widgets") and add it to all 8 belts one at a time, or combine it with the lane of 32 next to it, since all total that would be 31/s ore put on the 30/s belt, which is fine imo) For me, balancers are all about the abstraction. I don't want specific production chains tied to specific miners. I would rather 1 production line using iron products draw from all iron sources equally. It's the difference between having everything slow down as a patch depletes vs having entire sections stop, potentially causing the entire base to stop. On top of that, I don't want to care that a specific row has 36 miners and comes out at a rate of 23.4/s. I just make sure there isn't a far bigger (You don't seem to want to discuss trains, but I'm gonna add this in, since trains are the highest frequency use-case for balancers. Trains don't add "variability" if used properly. Using them the way you described does. But given a proper buffer and enough trains, there is no "travel time". A train should leave with items in the buffer and a new one should come in before the buffer depletes. This gives a very consistent output rate without the hassle of routing belts everywhere. It also allows for abstraction between ore patches, not just within.)
Wait but, your idea here as you are saying is to get rid of variability, but your method leads to the base having variability as the miners turn off at about the same rate that the balancers do if not maybe slightly faster since they cannot help for/account for each other. Though your way uses less resources in avoiding the balancer. Also you can use a balancer that goes from higher inputs to lower outputs if you want to guarantee the consistency that you are looking for ^-^ Also if you use blue belts on the miners you could split it off into multiple red belts to really get the max mining at all times I just realized.
Interesting video. But this hasn't resolved the issue of depletion, it's only delayed how long it takes for you to feel the effects. You'll still have miners that dry out and slow down the production of random components in your factory. It's very interesting that you point out how trains introduce variability in flow rates, since that's precisely where most people are using belt balancers. But I think you have the logic backwards. Trains are an additional source of variability, but they also are the best way to solve the main problem, which is miners that don't remain consistent. Trains allow you to dramatically overbuild mining outposts, which solves the problem at a different level. Use balancers to evenly extract from multiple mining outposts at the maximum rate locally, and overbuild mining outposts with train systems to ensure a consistent flow globally. Using maximum extraction rates locally guarantees that your trains remain balanced, while giving you the largest amount of buffer to keep flow rates stable at a global level. The best part is, you can measure the entire system globally to ensure that you've got ample time to be alerted and set up new outposts before your excess capacity leads to down time inside the factory. Use a giant train stacker to simultaneously buffer full trains, and measure how many full trains you have using circuit networks. If the number of full trains starts dropping below some critical level because trains are waiting idle at outposts to pick up more ore, you know it's time to build more outposts. With your system, there's no builtin way to realize that miners have depleted, without noticing low production at some part of your factory. You could measure the belts directly with circuit networks, but by the time you hear an alarm, it's already too late to add more miners to get 100% flow rate. Plus, I don't see how you handle ore patches that are so depleted that they can't supply a single belt. At some point you have to start combining the output of multiple mining patches. Trains are just a really convenient way to do that. Trains solve the problem at a higher level, by taking advantage of their capacity to move far more material than belts. While you could still dramatically overbuild mining capacity without proper belt balancers, the miners will deplete highly unevenly because miners stay inactive when the belts are full, which leads to the same resource patch becoming irregular faster than if you're using balancers and trains to buffer the output. At the diagram at 10:00, the middle of the blue miners will take dramatically longer to mine up all the ore than all of the combined purple miners will, because the middle blue miners aren't going to start working until some of the other blue miners deplete, but they're sitting on the most sense patch of ore. If you built two similar mines and used balancers to evenly deplete the ore patch, and chests and trains to buffer the output, those two outposts would produce an even flow rate for much longer than two of your outposts would, because of uneven miner depletion. Still, this is all pedantic. There's no "wrong" way to build a factory. People love watching DoshDoshington because he often does crazy and weird things that look stupid, because he's playing by the rules of some nonsense challenges. I enjoy watching his challenges because each one of his bases is usually much different, because he hates doing the same thing twice. People love your factories because they're all very regular, almost like mathematical art, and there's a sense of predictability to everything. Some people like large trains, some people like smaller trains. Some people don't even use trains at all, and prefer to use belts. It all comes down to personal preference.
As an aside, most people build train dropoff points so they supply an even belt of output. So if you know you're depleting all belts evenly, you don't need to use belt balancers on the output side, only on the input side where the miners are. Then again, I've been playing around with a lot of different train unloading station designs, because it turns out that it's really hard to design a good unloading station, and most designs have some pretty serious flaws. It's non-trivial to build a train unloading station that can output a fully compressed belt, while evenly unloading all the chests. Many of the trivial circuit controls for even train unloading will fail if used in non-optimal conditions, or with different tech levels. That's likely why most people just build a balancer right after their train unloader, because it solves most of the design flaws with train unloading. Plus, managing train networks can be scary, with the potential for deadlocks and all sorts of other chaotic problems. After playing around with trains for a long time, I definitely understand why some people prefer to use belts. To properly manage trains at scale requires some decent circuit knowledge and intimate knowledge of train routing. Which is likely why the new Factorio 2.0 update will have some pretty big improvements to trains and train automation, making the barrier to entry a lot lower to get into advanced train stuff. Trains are crazy complex for being such an early green science tech.
5:30 ...that's belt balancing. You're just doing it on the design side, which requires a very aesthetically unpleasant spaghetti of random belts connecting one or two machines here with one group waaaaay over there, rather than on the transportation side after you leave the mine. Imagine if you have two separate mines. Personally, I'd say "minimize the belts used to get the raw ore to the factory, then balance it so I can build a grid of equal size." If you decide to add belts to your design, you have to potentially entirely overhaul your belt network. If you balance, you just add another saturated belt. I don't even play factorio - I play satisfactory, but the idea is the same. It's kind of like the question of manifolds vs load balancing. Factorio doesn't have that choice as much, but would you run a belt to each machine because it guarantees throughout? Nope. Manifolds are waaaay simpler to design, build, and expand. As for all the stuff you're describing at 7:00, just reduce the post-balanced factory size by 20 or 30%. Or let it peak output early and slowly peter out, maybe saying "this section will be permanent but after the miners deplete, everything after this market will be starved so it can go"
But isn't the role of buffers to even out the flow with variable inputs? Train might deliver it's cargo in burst but the overall length of time it takes for a round trip is roughly the same. Thus you use buffers to accept all of it's cargo as fast as you can which you can then slowly distribute at even flow rate to the destination. Only issue is when your buffers become empty but all that means you need to increase input.
I'm skeptical of even finishing the video after the disaster of this failure in logic. Happy to see someone else pointed it out in the comments before I did. It's truly baffling how a person with the ability to play this game to such an advanced level can miss the idea of buffering. I'll hear out the argument about balancers but I do wonder if this guy could play more advanced mods with the stubborn attitude towards learning trains.
It's probably less of a concern in the very late game with very rich ore patches and high mining productivity, but no matter what you do, the ore patch _will_ deplete. You _will_ need to eventually build more mining outposts eventually and your existing patches won't be able to support the desired throughput. This is why I generally don't factor in ore rate into my calculations and just try to mine more than I need. There's also the matter of an uneven _consumption_ rate. Ideally, consumption rate is much more easy to control than ore production rate, but it is still possible to have variable consumption rate. The most common example is in the mall. You generally don't consume resources making building that aren't being placed, and the buildings will only get made, and hence consume resources, when you're actively expanding the factory, or when it's building back up the buffer consumed in a recent expansion. There's also the possibility of uneven usage in the science path, but there using a balancer is little more than a way to hide the problem rather than an actual solution. The actual solution is to fix your ratios and builds.
I'll be honest here, I will get bored of a factory pretty quickly after I've finished building it. I want enough mining production without depletion to be able to test the SPM for a few hours but after that I'm done and I'm onto the next factory. I've done the outposts, I've sunk the time into doing all of that and I don't find it fun. I've never used belt balancers ever, so to be honest again, I don't really even know if what I'm saying here completely baulks the argument I make in the video. The thing you say about variability in the shopping mall is absolutely true but I see the issue differently. I see my factories having two stages; 1. construction, and 2. SPM testing. I won't begin stage 2 until every pipe and every conveyor belt is completely full of ingredients. If ingredients are still being allotted to conveyor belts then that is taking resources away from science, as you say. If we are still extracting resources from the factory in order to build it then any test is pointless. We are still in the construction stage. So, yes, that variability exists for the factory, but it shouldn't exist at the moment of testing.
What kind of SPM do you shoot for in your factories may I ask? I play stock settings on vanilla and I have about 30 outposts in my 600SPM factory. I can't imagine relying on just the ore patches in proximity. Another way to play is to grow the factory over time. For me construction is ongoing so there is no construction stage and science stage. I started out around 120 SPM, built up to 240, and then went to 600. Next I might go to 1000SPM. I use trains for EVERYTHING and there is no variability in the flow rates into each section of the factory because the trains are over built and they buffer so the input belts are always 100% saturated. There are so many ways to play this game, like you said, and no right way. This is just my way for my current playthrough (on my 6th playthrough of the game, all the times before I used a main bus base but never got beyond maybe 120 SPM) @@theotherbigfoot
this woud be a problem for me for like the first 1-2 hours of a playthrou. The moment i have Trains, 100% of my new ore patches are coming in by train and if i don't balance the belts bevore going into the buffer chest, they will forever be uneven and the train pickup time will be limited by those slower filling chests
Balance the chests logically instead of using splitters, works better. Also it will even itself out if one chest is more full than the others. Calculate the average content of every chest with one combinator and disable the inserters if their own chests content is larger than the average ammount.
While I understand your point, not using belt balancers at the volatile input of the system, be it trains or mining drills is kind of missing that point. Use a 7 to 4 balancer so you have a filled 4 belts for a long time and you can go from there with the 4 belts of ore output. Using them within the factory instead of just at the inputs is a crutch to have stuff going, even when throughput is insufficient, but that's okay, since things will eventually back up and it will naturally hit its maximum throughput. It's not the cleanest solution, but it is a solution nonetheless.
There is nothing to really understand in his point since its wrong... His max throughtput works because he did a 8:5 compression to a garanteed saturation. He could have achieved the same result with a 8:5 balancer without taking half a day to create clusters (which are also variable at some point, and which he will need to adjust every 2 hours anyway all around the map...)
i found that the best way to treat ore patch is to set to output north and south(east or west) so that the first ore to be mined is the center part. by the time the center most part is deplated, you will have consumed the most out of the patch so you basically obtain the max output for longer
Ballencers are like buffers, unneeded in most places, but critical in the few places they are needed. Normally just with trains, and things that are basically trains, however some mods do add other things with odd pulses that this also applies to as well. A ballencer is to make sure things load and unload evenly into buffers, that exist just to eat the variable flow of trains. Best use is at a mine, where you set it up so a train will only show if there is enough to fill it, so its best to make sure each wagon has a buffer, and each buffer gets filled at the same rate. If you design your base correctly, you can prevent the need for balancers on the unloaders by designing around belts as the unit of consumption, but otherwise a balancer is useful to make sure the 3rd car isn't drained at half the rate. Otherwise they are unnecessary, as the large numbers of trains, and well chosen points to balance, make them unneeded.
Sorry if I'm missing something obvious, but why not just set up an ore buffer near the smelting columns, and send out trains to mining outposts as needed to top off that ore buffer. The trains then just have a condition to stay until emptied. Do the same thing at the outposts to minimize unnecessary downtime for the miners (if you have something larger than chests, maybe even enough to empty the patch), and you have a constant rate without carefully sectioning off your miners. Especially since they will start being depleted eventually anyway.
That's called 'factorio basics' and most amateurs have that figured out. *Baby's first standardization* is an inserter-chest-inserter setup going in and out of rail cars.
Personally I use balancers on my ore patches more to compress the input down. Sure the ore patch may have 8 lanes of mining drills, but some of those will have less output and will deplete sooner, so typically I will compress it down to 4 belts. If I need more than 4 belts I'll go find another ore patch; there's plenty of them! With a well-designed train system you can reasonably reliably know how long a train journey will take with only occasional traffic issues (and this'll be even easier with the expansion's elevated rails!) and use this to judge a suitable buffer size to compensate for the intermittent times. I very rarely use balancers outside of the mining outposts. Might use one or two in my main bus just to rebalance or compress belts down as resources are consumed (for example I'll compress my 8 copper belts down to 4 after the green circuit section) but otherwise I just rely on the fact that balanced belts in = balanced belts out! A full train unloading 4 belts of materials will unload the same amount of materials onto each belt, so there's no need to ever balance that. The output of that block will likewise be evenly split. Just... don't mess with it by hand-feeding some machines XD
balance is not about throughput. For example, if you have one line of green circuit, and want seperate flow for few items. With balancer you can make that, without wait each (one by one). It's not a big problem, but at start not too much resources (but there is a lot to be built). Main use is trains. Most problem load/unload wagons/chests (>4) . In one chest zero items, others full, efficient lost. On this way need not only belt balancer, but belt-each-line balancer. But in vanilla it's not that necessary. Only for big_mods/megabases it gives you more time, before production will slow down, and you should personnaly check what/where happens. With overproduction it's still has no meaning, but only you design your base, and deside how it would work
For me, i only use balancer to fill my ore train so when a train is stuck i know exactly that it is because the patch is totaly empty.Then i stock those train near my smelter zone in a full-of-ore-train buffer, to make sure travel time does not matter. And insiste my factory, no balancer, only perfect ratio as far as i can, nice video tanks a lot
I absolutely LOVE how the end message for orefield setups regarding "fixed, guaranteed flow" is literally "it's basically gut feel." I totally get what you're saying. When I first played the game, I remember seeing posts regarding the maxim, "overproduce far upstream, it's way too stressful when you run out of source materials." I've always found that true, for better or for worst. It can be super rewarding to play with absolute ratios 100% of the time, but I still have the most fun in factories where I suck up everything I can and work out the magic in the middle of the base.
You have point with the flow rate mindset, but only some belt balancer are "needed" because of variable input and/or output flow rates. Most players use trains to transport ore into the base. The loading and unloading trains evenly is needed to ensure a constant flow rate. (or one cargo wagon will empty and the other are not) Like you said, trains have variable "flow rate", but trains are a method for abstraction and modules in your factory. Sure you can design a factory that create x science and calculate all the necessary inputs/miners/... , but it is much harder than building a factory and use some belts balancer and increase the input if you need it. Using belt balancer to much can be a indicater that someone missed the idea of ratios and can make the factory more complicated than needed.
You can't keep the same flow rate belting from a single ore patch, it will always dry up eventually if you play long enough. This is the beauty of trains for me, taking resources and distributing to where things need it, which acts as a sort of balancer. I used to rely a lot on balancers but they are kinda big and clunky, also things will balance themselves out if you let outputs fill up. I just priority flow which ensures the whole belt gets used across the factory.
i really like your outlook. Everyone plays differently and you couldn't be more right. Like I use balancers everywhere, like an addiction. You use these precise rates but my method is nothing like it. I prefer to make a surplus of everything, if im not filling buffers, I'm not producing enough. Sure I need more power than a roided out strong man, but thats the way I like it. I end up wiht a similar result, just in more space. Youre factory looks incredible, your style is how a feel like a city block factory is meant to be made. Clean, efficient, and perfect. Love your perspective man
While balancing around maximum output has its problems I think in case of mining operations it is worth using balancers but with shrinking. I calculate the amount of drill output I get, deduct 20% of it and then create a balancer from however many lanes I've got to the 80% theoretical maximum throughput. This way in case of bigger patches they will last a long time, especially if more levels of mining productivity are researched.
belt balancers are eliminate variable flows. fun video! edit: if you want a fixed flow rate for as long as possible from each mining outpost, the only solution would be to balance the miners so they all input onto a single yellow belt.
I might be a bit too new to the game, but this just does not make sense in any way. how does this balance your output, when this doesn't do basically anything? You get basically the same thing, but better with balancers. What am I missing here?
The problem with building extra miners per belt is that miners that aren't mining are wasting a small amount of electricity just by existing. When you put the ore in a train, it all equalizes out anyways, as long as you have a queue of trains coming in, your throughout is uniform.
In order to even out the output you can build a capacitor. First batch of ore is collected and put into chests. The belt gets full. When the belts start to dry out, the excess of ore collected at the beginning is now being put back on the belt
just crazy idea. to keep a fixed flow for as long as possible, wouldn't it be better to try to prioritize the mining from the central patches since they have a higher richness than the outer ones?, so to try and mine the ore as evenly as possible to avoid the "outer edges running out" which would drop the mine output
I generally don't use a balancer for maximum flow at a mining outpost. I usually use it to just spread 'load' across the patch, and to get the number of belts down to a number that makes sense for my train size.
❤❤❤ I think I love you! You are essentially cutting the maximum ore output from a patch in order to provide a longer period of consistent ore flow rate per belt. This means you might have 1 or 2 fewer output belts but your system is more predictable. It also means you are constraining your factory bottlenecks to the input and forcing the input to never exceed the maximum throughout that the system can support, and also buffering that input so that the system is less likely to be starved later on. You know, one of the biggest problems I think I have is the mall idea because it encourages segregating materials for construction from materials for science and well as limiting the expansion and scalability of these items. Not to mention encouraging building along the main bus rather than formalising the idea of sub buses which are tailored to specific purposes. I suspect I know the answer, but have you read "the goal"? I'd be fascinated to hear your impressions of that book.
If you design with factory-wide guaranteed flow you create these issues: - every expansion requires full vertical implementation - inability to prioritize production - fragility - any belt going “slack” will ripple through all the production downstream, worst case scenario blocking entire production lines.
Trains also provide guaranteed flow. If one train can't being enough resources to the unload station to guarantee a constant flow, you can just add more trains to the same unload station.
This is a very interesting way of thinking. I like it because you can design a factory perfectly to maximize every resource gathered. Where this falls appart though is the moment you step into multiplayer 😂
What about equalizing both sides of the belt? If you don't do that, one side of the drills might empty much faster than the other, leaving the remaining half of drills outputting down to half the original rate.
I don't want to be pedantic in return, and I don't want to say my friends are clowns, but like, we play together on a server and love it! But I think your argument isn't so much are they good, but are they precise We don't sync up between plans so sometimes there's a miscommunication or misunderstanding or even just sometimes bad math because someone's baby is crying when the graphical calculators come out. Not to mention we play on different timezones so sometimes we're fundamentally in a place where things don't run as smoothly... Balancers allow us to just shore up supply, in a hodge podge spaghetti build that's gotten out of hand, despite the best of intentions, simply because there was an issue somewhere when we were planning weeks ago. Balancers alleviate the problem and while the impact is less production, we get more across the board than we would if one line was 100% fed and the rest aren't, which of course creates the need for balancers further down the road and the problem grows... But that's the nature of precision Vs good enough for some nerds being nerds to our hearts content in our spare time
I will accept that I did not consider the multiplayer angle. Fair. Besides I'm not saying they are wrong, just why I don't use them, and I don't play multiplayer so it makes sense. I am also pedantic and I like to be precise. Thinking about it, I don't think I would enjoy multiplayer.
@@theotherbigfoot I get you're not saying they are wrong, but that when building mega factories, we can do better, which is what you lay out and you're right. It just won't work for me and my friends and how being human gets in the way of us playing Great video btw :)
Two mods that help a great deal: Miniloaders and Merging Chests. The second mod lets you slap down arrays of steel chests and then merge them into giant chests, warehouses and storage depots, and using miniloaders, it makes it easy to flow belts into and out of these large storage sites. The miniloaders pulling on the contents of these storage sites is done evenly. Thus an even output is obtained. I use these for my mining sites to give me a bunch of uniform, compressed belts of ore. The sites are also nice because they can double as in line buffers.
In the miner example why would you not just feed the 6 lines into a balancer and only pull 4 out this gives you a buffer and you can even setup logic to tell you when production is slowing down this will make sure the variability does not trickle down allowing you some time to setup another mine I personally don't use balancers, out side of miners, either I use belt compressors compressing the flow down to as few belts as possible and then splitting off the most compressed belt for production this make sure each section of the base gets full production and then adding more plate production if required
with the addition of the concept of buffers, it all depends on your expansion rate if you can keep up with your factory consumption rates. You could also just do on-site smelting to lighten the logistic load
Imo your clumping of miners is just a temporary solution. It guarantees full belts until the patch dwindles. Topping up such a belt is arduous. I like to think of trains as that guarantee. I just need the sum of all my mining outposts production to be greater yhan my consumption. It doesn't matter if an outpost is dying and produces just 12 per second. Its about the total. That's where balancers come in: loading and unloading trains evenly. If you don't guarantee an even unloading in particular, you may not have saturated belts. You will if you balance it, as the wagons are drained equally and thus the set amount of inserters is always all active. I see no other way to guarantee saturated belts and guaranteed constant flow. Mines are by nature unable to provide that. Only once you use trains to bring it together you can guarantee constant full belts, no matter which mine produces how much at a given time. Maybe i just don't understand what you're doing, but to me those are all temporary setups. Sure you double up on the miners, pushing the issue into the future, but it'll eventually creep up anyway. Am i missing something?
The simplest and best method that works for me is to have a balancer of input greater than the output, 6 - 4 for example. Then you just mix your outer most lanes into one to improve their longevity. By the time those 6 inputs can't feed your 4 outputs then the mine is getting pretty close to being drained. you might need to rejig the inputs a little but you're also at the point of getting ready to look for a new patch of ore if you haven't already. always over supply the inputs to the required output. My big train unloaders usually have 3-2 balancers from each side of the same train for compression.
Your argument is based on the idea that maximising the output of the mines (or other resources) into the factory (or its segments) directly will improve the output of your factory. This is an incorrect argument. The goal you want to achieve and the method of input are disconnected from each other. Balancing or not, you can achieve a fixed output, even introducing trains to the equation. The goal is to maintain a buffer so you have a fixed input into your factory. In doing so, you ensure that your mathematical perfection persists, because you know as long as you have a fully packed belt (or two, or four, or...) coming into your factory, you'll have a fully packed belt going out with whatever item you wanted. Allow me to explain. But first: - We'll ignore the complexity of the factory's internals itself, because that's a mathematical nature we don't have to address. We'll assume your ratios are correct. And with a compressed belt you'll achieve optimal output ratios for whatever you're intending to make, so ultimately that's the problem to solve. - For the purpose of this counterpoint the definitions of "raw resources" and "resources" are synonymous as well. Copper wires are to green circuits as copper is to smelters as copper plates are to assemblers. - Finally, a "construction unit" or just "unit" is any segment of your factory which consumes resources to produce something else. Doesn't matter if you have a unit that is exclusively copper ore which exports copper plates elsewhere or a self contained science unit which takes in all manner of raw resources and combines it on site into science for export. --------- The assumption that you can ever feed a consistently full belt of resources into a unit directly from a source is wrong. Flow is always variable directly from a source, even if it's just a ripple effect from depleting resources in some fashion or another earlier in the chain or an overabundance of resources clogging up your current unit because it mistakenly assumes it can clear enough materials to make and deal with byproducts (i.e. ye olde oil processing and dealing with multiple oil types and petroleum). Maybe you add another unit that consumes more of a particular type of resource (like another thing that requires green circuits but don't get around to making more green circuits to accommodate). Maybe you don't notice the gradually falling off oil output from a field getting to its minimum and things progress as you expect until they don't and bottlenecks form that require you to start tracing backward. The goal is to maintain the full and complete packing of a belt (or series of belts) heading into a unit. That is all. To properly achieve this, you must use a buffer, and that buffer must (net) fill faster than the unit consumes it. Regardless of the potential burst output of a miner array or unit, as long as the input into another unit is always full, you've achieve YOUR stated goal. This is where direct feeding versus belt balancing is up for debate. I'm in favour of belt balancing because it ensures a number of conditions are met: 1. When input is substantial enough, there will never be a blockage because the balancer ensures that surplus goes to alternate paths. 2. When input is insufficient, you will still get output from your unit, starting with a gradual and equal reduction across your unit down to a slow trickle from early portions of smelters or assemblers, ensuring every belt has functional components at least. 3. It also ensures that no one source will inhibit the input or output of your unit. For example: You have 4 belts leading into parts of a unit. For inexplicable reasons, one path consumes resources faster than another (maybe you miscounted a number of smelters, maybe you partially upgraded belts/inserters on one path but not the others, etc.). If the line going into that path wanes, it will produce less resources than all the other paths overall, especially when it consumes all the resources it needs on its path and becomes deadlocked while every other path chugs on merrily. Belt balancers aren't always necessary though. What's important is making sure that the inputs to buffers are properly balanced. If you can do that without a balancer, power to you. If you DO use balancers though you can ensure that trains are filled equally and optimally in the shortest period of time. If you think of trains as a type of buffer, then you want to ensure that each car is filled as fast as possible to get that buffer to your units as fast as possible. Trains are "burst buffers", large-ish packets of resources which are meant to be offloaded quickly into local chests (another buffer) and then swapped for another train. The idea is you have an excess of trains supplying the resources your unit needs to function in more than sufficient quantity so that when the time comes that a ripple occurs, your factory doesn't notice it and it stops at the level where the ripple originated. At SOME point, you need to balance your resources to ensure an equal distribution of resources for your units. If your balance is just stuffing a train full of resources and then pulling it off of the train at its destination into chests, that's balance enough. tl;dr: Balancers aren't important, maintaining a buffer is. Balancers can ensure an equal distribution of resources across your buffers. But merely stuffing as much as you can into the buffer in as short a time as possible and ensuring the buffer does not deplete is enough to ensure full compression. Balancers beyond your buffers just ensure optimal use of factory construction units regardless of whether a segment of your buffer is depleted early or not.
That chaos added by trains is actually good! Your ptch started to run dry? Add another patch with the same name! This way both of them whould overproduce and only send trains when they are full, they provide constant flow of resources
There is always a bottleneck, for me that bottleneck is the belt, and I have no problem with putting too much on a belt and letting that balance things out.
A lot of really long comments in the comment section :D here's my long comment: I use belt balancers often, but only for loading and unloading train cargo. Like you said, "trains introduce variability to the flow", and one way you can reduce that variability is to place down chests next to train stations that act as "buffers". Items are placed into the buffer chests while the train is away, and when the train is present, stack/bulk inserters quickly transfer items from the buffer chests to the cargo wagon. If you want to maximize the speed at which the bulk inserters transfer the items from the buffer chests to the cargo wagon, the items need to be evenly distributed in the buffer chests. And the only way to evenly distribute these items evenly is with belt balancers.
what you need is intentional grouping of mines, which you seem to do already based on "feel" rather than analysis and intent... and a compression-balancer or reduction-balancer....something that takes multiple FULL lines and compresses down to a fewer number of lines (while also balancing lanes) such that, as the miners deplete, the output remains compressed. This becomes easier as mining productivity increases...and once it's to level 15+ becomes almost meaningless/pointless as the mining prod makes each line require fewer and fewer miners to compress a full belt....proper grouping and mixing/lane balancing still gets the job done. Also...use of trains at stations and proper balancing and buffer design/unload mixing will ensure your base has a max flow rate......if properly designed... and if you stay ahead of the rate that you deplete outposts. Part of the problem you face is you aren't building ore outposts frequently/soon enough to mitigate the effect of depleting ore fields.
My current strategy: If an ore patch has a maximum throughput of X number of belts, I'll throughput limit that patch to between 20-30%(rounded up or down to the nearest integer greater than zero) by compressing the belts. I like the slow burn.
I think this is very odd. You can get away with not using balancers inside the factory just by balancing your train cargo and then in the dropoff you know you'll have n lanes of fully saturated belts. You don't need to do weird stuff with mining patches to ensure saturated belts
I’ve never felt the need to belt balance, but I would only ever need it for ore… and I just smelt ore on site and train it back to base to maximize efficiency. I also move my base closer to the resources as the game goes on since distance is the largest killer of throughput. I tend to do a lot of buffering and using “belt wrappers” instead of balancers. I plan for belts to always haul maximum items and when they aren’t, I release buffers first or belt wrap a new source of the item. But, this maximized line is also why I tend to avoid overuse of “blocks” and “main bus”. Each line is treated as a full production line. Spare items that won’t be used on a full production line are rolled back into buffers or “belt wrappers” to fill back in on lines where production has dropped. But, I love the look of full belts even if they aren’t moving. I don’t like “balancing” the belt. I like it full.
Thank you for reasoning this through! I hadn't realised this was the solution to the problem i was having with flow rates - i.e. I was building my factory based on the original maximum rate. I really like your solution and will impliment it in my next factory for waaaay better control. Awesome man, thank you! :)
I have always struggled with this topic in Factorio. I'm definitely unreasonably pedantic myself, and belt balancing is one of the biggest road blocks I've had in Factorio. Definitely a sensitive topic though, hope you won't be getting too much 'feedback'. I hope I can give this challenge a fresh start with Space Age after watching some of your videos. We seem to think in very similar ways. Thanks for your content!
I just found your video and this is very interesting (and contrary to those vids that I've watched before, bo other content creators here), I am going to trust my gut and give you, my good man, a sub, and carry on watching.
3rd comment i know but holy crap i just realized you are building your factories exactly like i am! brace for long comment. i sought to modularize my factory by blueprinting linearly mapped (16x16, 32x32, 64x64, 128x128) 'grids' of production lines separated and connected in a repeating grid pattern bordered by rail lines. i treat these individual grids like monoids. meaning each grid does not depend on other grids existing in any certain order or priority or placement thus "associative" if you will allow the monoid analogy. then i can just haphazardly place these production line grids in any sequence that i wish. a depot is included where outputs are gathered by train, buffered or balanced, and sent out by train. i designed the lines such that trains can never be path blocked. there is always a path because each grid connection is the rail line itself in a repeating pattern. no proof but i intuit this is the case. whereas most players build fragile rube goldberg devices. a monoidal factory is the most robust factory one can build. drop a nuke on any random spot and the rest of the factory will still keep going. i stopped playing the game because i am obsessed with efficiency and robustness. so i sought to optimize grid production line space by figuring out the best configuration for parts in the grid that would pack the most information. so i set out to write a genetic algorithm to do it but my lack of technical understanding of how to optimize this packing problem has stalled me and ive been putting it off for years. thanks for jogging my memory!
The blue belt has 8 item room in one square, the steel chest has 48x50 or 100 item in stack. The blue belt can move 45 item/s to the next square. The green arm can round 864°/s, so it can round 360° in 0.4s and it moves 10 item to 2 square distance, so the speed is 2 square distance/0.4s*10 item = 48item/s to the next square.
I like this a lot, i think its basically what ive always done just smarter and more intentional. I don't quite understand the drill blocks yet but i think ill try it in that manner!
The only place I use balancers is for ore. I just make lines of drills and however many lines I end up with I just stamp down a balancer that compresses it down to the target amount of belts. Say I want 4 belts out, but I have 8 belts of ore, I just make an 8 to 4 balancer. In this way, the only time you see a reduced output is if the total operational miners aren’t sufficient.
Are you suggesting that using belt balancers WILL provide mathematical precision? Because I have a problem with that statement the same way you probably have a problem with mine.
As a beginner, I use balancers. I don't even know what exactly awaits in the late game. I don't have later item recipes unlocked and my planning ahead is based a lot on peoples recommendations. Started to build a main bus base after spending like 30 hours on the tutorial and the started base for free play. I just have 60 copper per second smelter and also 60 iron puss another 60 goes into steel. The mining is just slight overkill, so I could run the furnaces. I have no Idea how big I should build my next productions really. I don't know how much materials the later things will need. But it's fun to discover. I left some room for smelter expansion. But the idea behind the build was that someone in a video said that this amount is good for completing the game. But after I learn how the game works, I like the idea of more precision and neatness.
I find this argument really interesting. I tend to use balancers on mines, and then use trains from there to the factory, but from the unloading, I don't use balancers ever. It's all flow rate, and one belt of copper here, two belts of iron there, and so on. I guess the trains make it so the variability of the mines is kindof a moot point, whereas inside the factory, precision governs all. Thank you for reminding me!
Discrete values aren’t just a miracle. The main juice of trains that they can guarantee constant value every time they arrive. Problems like “there is not enough throughput in this station” lies in the field of “Build moar trains. Build moar miners. Build better upload/unload designs”. I agree, that in one production block u can (and i’d say) must build without any belt balancing (cuz all balance should be put into mathematics of ratios). But when I need to make my trains leave mining outpost every 20-30 seconds with 4-inf train carts full - i need to make all buffer chests sufficiently equipped in approximately same time. Belt balancers (or belt splitters in that case) are used. Bro basically sais: I dont want to balance output of whole patch of resources, I wanna balance it by dividing into some balanced pieces. Plus I don’t care about imperfections of ore distribution through the patch. Plus I wouldn’t think about how depletion of even one drill will affect whole scheme. “It’s most of the gut-feeling.” + “I want mathematical guarantees “. Schizophrenia.
Very interesting video with cool concepts. I always make more machines or miners that would over-saturate the belt. I always plan to make more than the output is capable of handling, this way I can be sure that, if I supply more than enough resources, I'll always get a belt's worth of products. I think my approach is very similar to yours, and it's something that really works. But the biggest issue is when consuming the belts, some can get less saturated than the other ones due to side-loading, imperfect ratios on recipes and other weird factorio cases. I don't use a lot of multi-belt balancers but I do use a lot of side-balancers
I appreciate your gameplay approach but notice some practical issues, especially with the fluctuating needs of your factory. Constantly monitoring each conveyor belt for specific processes, especially with raw materials, becomes impractical. My own experience led me from striving for utmost optimization to realizing that a simpler approach was more time-efficient. When playing in a team, it's unrealistic to expect everyone to adopt the same building style or to understand the intended use of each conveyor belt. While communication is vital, setting basic standards for things like directionality, the use of main buses, and robots is essential. Overly strict rules, like those regarding belt balancers, can backfire. Although your strategy may work well for solo play, it might not be as effective in a team setting. My gameplay has evolved to prioritize practicality, focusing on being quick and efficient, yet simple. This shift is similar to the reasons behind Python's popularity: its straightforwardness and readability, despite its performance limitations due to being an interpreted language.
okay, sure, you want a guaranteed flow in the factory, but as you pointed out the issue is at the mines because of uneven ore patches, which as you continue to play new ones get built farther from the factory. So use trains, and use belt balancers at the mines to evenly distribute ore into (the buffers that load into) the train cars. So it really doesn't matter what the maximum or optimal flow possible from extracted ore, what matters is that every train car gets loaded as efficiently as possible, which is why even distribution is what you want.
also, because inserters move resources faster between inventories (cargo wagons and chests) then they can pull or insert onto belts...the use of buffer chests at ore patches and at unloading stations....and proper splitting/balancing at BOTH ends is required to maximize compression on all belts in your factory. Skipping this is one reason why you may be forced into the mindset you've adopted.
I see your error. You said you dislike trains, and that's the same reason as balancers. You want every factory to run full output at all times, rather than calculate for max potential. *Buffer chests* are your best friend for that purpose. You say "I can connect one belt" well the rest of us can make 4 carloads and ship it to the other side of the world, and have a consistently full belt at the other end. You also seem to misunderstand what balancers are even for to begin wih. The MOST important use of balancers, is splitting ore belts into trains. So that every car loads X stacks, instead of 7 stacks in one and 3 in another. Or going into smelting, you don't even WANT 8-8, 6-6, 12-12, etc. you want 8-4, 6-8, etc. splitting input equally between production facilities.
As an experienced factorio player, this is lot of blabla just to say: I love doing a lot more work than necessary Usual process: slap a lot of mining drills on a patch, balance it into chests, load into train, unload, done, now you can focus on better things than micromanaging drills Oh and a patch runs out? Add another patch to the train network :)
I have a similar mindset myself, tho I use trains so the reason is different (I want train to be loaded as fast and equally as possible so they enter and leave stations as possible). But I goes on to saturate my belts from ore patch differently: I use priority splitters, in the example case I would also take 4 (maybe 3) belt out of the patch, and just load as much as possible from the 6 row of miners into 4 lanes, it's easier to setup and make identifying dying outpost quicker (one belt is obviously empty).
trains have higher bandwidth but also introduce latency. ideally when a train leaves dropoff station because it's empty it would be immediately replaced by a new full one that is right behind it, this plus the chests as a buffer should eliminate variability. on the mining side, build more mines than you need, so each mine works at full speed for half the time and then waits for a train. then as their output declines other mines will pick up the slack. basically just make sure mines have the ability to mine more than you need and then bottleneck them, if not build more ahead of time. I haven't played in over a year btw, so maybe this only works with circuit networks that set number of trains per station based on how much ore is in the chests. hope this makes sense.
Honestly I don't agree on the hate that being showed here in the comments. I found the video very entertaining and I like your logic. I'm in no way a Factorio veteran, but I like the game as a casual, and I found myself always using belt baencers everywhere in my ore patches and every time there was a problem within a few hours due to less output, the very problem you pointed out in your video. I think the next time I'm going to play, I'm gonna think more long term and try to do something different, like you did. Great video!
what i personally like to do, is to use trains from ore patches, to a central smelting station, and hooking up speakers to the ore loaders, so that when the miners can't keep up with the trains, aka more ore is being smelted than mined, i have time to make a new mining outpost. Usually i have about 3 mines for iron and copper at any given time, and with this method i constantly have a full smelting setup, and reliable plate output. IMO your method requires way too much thinking, for a problem that could be solved easier, however, nothing wrong with either method. And yes to the guy that's about to post "plates stack 2x ores, so moving ores in trains are ineffecient", go do your own thing:)
Seems like a hassle. I just use two blueprints to mine ore, the mining half and the train half. The first just throws down a massive square of all the miners that will fit, power, roboports, beacons, and an excess of belts. I literally don't care if massive lengths of belts/beacons/power connect to nothing, it isn't worth the time to trim it once your factory is big enough to crap out construction materials faster than you can spend them. Then I have a second blueprint containing a train stop (pre-named as an ore type), train queue, buffering and loading on both sides of the track, a balancer, power, roboports, and a radar. I can just shove down multiples of the mining blueprint and a train blueprint (pick the right one to avoid having to re-name the station) and most of the work is done. I even have a generic mixed ore variant for the overlapping patches. Then just lazily connect the belts between each blueprint. Really, the hardest part is making sure that really remote mining outposts have separate logistics networks supplied by trains to prevent bots from getting stuck. It doesn't have to be some perfect ratio, because it's fast and easy to just build more mining outposts. You still want the balancer to make sure trains are loaded evenly and not sitting there waiting for miners to work on the last car, or never finishing loading it at all. I've sometimes added circuit conditions and alarms to notify me when the patch finally dries up so I can pull most of it back up again at my leisure. You can have the train stop disabled when the buffers aren't full enough to fill a train, but alternatively you can just have a network with plenty of trains and train queues at both ends. Far into repeated research this setup gets replaced by direct to train mining, but I usually play for mod content and not repeating research. tldr: You can't perfectly balance everything in Factorio, and if you try, my unbalanced factory will be bigger given the same amount of time anyways. Just quickly blast biters using spidertrons/artillery, expand/build more networks (power, tracks, logistics) and defenses, and quickly slap down extra mining outposts and ore trains. Have an excess of early production and taper off towards end products and science output. The alternative is factories stalling due to missing ore, which we would call a skill issue.
Initial setup: 8 to 8 lanes balancer "no balancer" setup: 5 lanes hear me out... 8 to 5 balancer. Guaranteed flowrate AND balanced mining through all fields.
2 things really. 1. Is a 8x5 balancer really possible? A splitter has two outputs so in my mind a belt balancer is only really possible in multiples of two. 8x6 and 8x4 I understand, but not 8x5. Can one be jerry rigged? Yes. Is it possible to build a perfect 8x5 balancer? Not in my opinion. 2. How many belt balancers is one expected to lug around? There are a lot of different configurations of belt balancers out there, either I remember them all by heart or I have a blueprint book filled with dozens of the things. I resent having to carry around all my perimeter wall segments and there's only 8 of those. It definitely feels to me like belt balancers are one of those things where if you do use them then you will find opportunities everywhere, but if you don't, then you'll find good reasons not to. There are compelling cases for both sides. Nobody is wrong but everybody is stubborn!!
@@theotherbigfoot1. Google is your friend, 2. I don't see the point, blueprint books can be stashed in the blueprint library and then you don't have to worry about it, I think balancers are a great tool especially with late game factories that use busses with multiple lanes of a single resource to satisfy the high demand that infinite research requires, it means you don't have to terminate lanes or use undergrounds to access most of your throughput
You can just use buffers to have flow garantees on your train outputs. You can have guaranteed flow rates by invreasing your extraction when you are running low too... No need to complicate things that much for guaranteeing flow rates.
I use belt balancers simply to avoid having to spend extra time planning around ore patches, which by their very design, are random. Even if I only save 5 minutes at each extraction facility, that adds up to hours worth of time saved pretty quickly. For me, I care more about the time efficacy that belt balancers bring.
If you don' like variable flows, another solution would be to use a cascade of splitters with priorities always to the same side, so you get a bunch of belts with guaranteed flow, plus a few more belts "extra" just to fill the gaps. This way you don't need to "balance" the amount of miners for each belt ;-)
Cool take, mate. I can't be bothered with perfect rations in my factories, especially when playing modded, but I often use something like 8 to 4 balancers for train loading, limiting the thoughput, but making sure that the particular mine will have constant output for a long period of time. I feel that's kind of what you are trying to achieve, but balancers reducing the number of belts seem better to me than combining mines manually.
This isn’t to say that your method is wrong, it isn’t but belt balancers can be blueprinted and provide an easy solution to a problem that is fundamentally unbalanced to begin with. In order to use your method we would have to calculate how many drills we need and add 20 percent on top to account for losses over time. And as the ore patch depletes the amount of ore will run out anyways. The only time this seems like it could apply would be if you go absurdly far away from spawn where the ore patches are all but guaranteed to run out. Only then would I see a practical reason to do this. For spawn or nearby spawn it would just make more sense to use a belt balancer
I get what your saying, but through the use of balancers you can achieve a guaranteed throughput. You do have a unique way of connecting your miners which makes more sense I think.
You can get the configuration Like that that you have for example 8 lines coming Out and you Balance Them to 4 so If one Line Runs Out you still will have 4 full Lines
Overbuild and multiplex the supply that enters the start of the process. Been playing Sea Block and that's so interconnected with feedforward and feedback loops that no real attempt at math can work for me. In SB the consumption creates needed byproducts used in supply as well as side products leaving the only viable distribution method to be oversupply. I'm not even close to worrying about belt saturation or balance.
I tend to just push the ore as far to one side of the set of belts as possible. build each furnace stack for exactly one belt. doesn't cause idling miners on mining prod. bonuses, and isn't too hard to patch new input/outputs, and will pretty gracefully handle any in/out mismatch. It is pretty splitter-heavy, but if I'm dealing with enough ore that it's a concern, I have enough ore that it isn't a concern. Scales pretty nicely for high science cost multiplier
I'm pretty sure that if each of us made this exact same video, whether for belt balancers or against them, each of those videos would be completely different even if the videos were meant to agree with each other. It's one of those issues that has an enormous variety of opinions. This is just my take on the subject. It's only really relevant because I need to refer to it in another video I am working on.
@@theotherbigfoot the insane variety of valid approaches is one of my favourite things about Factorio! I love learning the ways other people approach the game, and is one of the things that really makes me enjoy (and occasionally absolutely despise) multiplayer :3
I do the same for busses, always push everything to one side so that when variability does happen you know how it will happen. But I still use belt balancers in other places, especially in train loading/unloading stations, because I often want to minimize the amount of time a station is occupied by any given train. And lane balancers (scrambling widgets) are important in lots of places.
@@theotherbigfoot I agree on that :) The good thing is that we can play Factorio however we want! Did you play Seablock? Since you have to create the ores and not mine them you'd have full control over virtually everything in your factory.
@@randomguy-no5vh Absolutely agree. I've been reading through the comments as they arrive, everyone has a completely different take on the subject. Its awesome. I briefly looked at seablock a year or so ago but for whatever reason I didn't pursue it. It's not on my list I'm afraid. The approx plan is K2SE (which I've just started), DLC and then py. That should cover the next year or so.
I tend to slap down miners so that the entire patch is covered and just use a widget to get it it down to 4 belts that then toss into boxes. There are only 2 goals. 1) that the entire patch will eventually be mined and 2) that the train can stop and get full cargo quickly due to plenty of ore waiting to be loaded.. I do like the green circuit thing but seeing copper wire on a belt gives me the willies. Im a direct insert guy as much as I can possibly do it. In order are Direct Insert > Belts > Trains >>>>> Bots. I find the bots boring.
I'd rather just consume a belt or lane and shift the remaining lanes over. Better compression, and the builds are already ordered by requirements, so it won't affect output.
I think the idea is to overbuild the mines, make a balancer that fit your "fixed ratio" design, and basically not worry about it until the mine is depleting at that point you should already be feeding it from different source (usually trains) or the ratio go down, the fastest/close-to-automation way to build a mine and forget about it, even with your design the mines will eventually go down and the ratio go down. only when you have less mines to saturate the belts balancers would be unnecessary. And for the second half with the input/output blocks (city blocks without trains) this is what happen when you move away of the main-buss-design public meta. which is a balancers hive. I agree on play however you find it fun. And I find your factory blocks fun to look at. they look like beautiful Art. personal though, Unless you are playing in marathon, Going mega base or playing in a mod overhaul that push you to build a mega base, most those design patterns are just overkill/just-for-fun/Art.
Why build a factory based on its input and not its output? My approach would be to achieve a specific amount of output produced and to ensure that there is an excess amount of input to feed it. Excess input only really comes at the cost of power, and even then not always as things will stop when maximum throughput is achieved. In the case of trains, I use buffered storage to smooth out the variability and ensure that there is enough throughput to maintain a consistent flow of resources to the factory.
I agree with the mindset of designing for guaranteed flow, but I completely disagree with the idea that 'avoiding balancers' is the way to achieve that. The balanced belt example at 2:50 gives 8 belts of variable output. Your method at 5:00 to manually cluster and balance etc, gives 5 belts of guaranteed flow. It's not the fact that you avoided the balancers that's important, it's the fact that you're only trying to extract 5 belts of output.
If you had just used an "8 to 5" balancer to compress on the first setup you would get the same 5 belts of guaranteed flow. Less manual work trying to cluster things. Less worry about lane 1 having too many edge miners running out. No 'Spare' miners that aren't capable of outputting.
Using the balancer is less work with more time of guaranteed flow because you'd only fall below 5 belts of output when the entire ore patch is incapable of providing the 5 belts of output. In your manual setup you could have belt 1 running at 95% throughput because it happened to have two edge miners that stopped earlier than the rest. Especially because in your example the 'spare' miners you don't use could have been looped into the balancer to provide input when others start failing.
Not strictly correct for the last part, you still can have less than maximum output if you are limited by the throughput of a single lane. Also, left and right lanes essentially work as two whole separate networks, so you can have one runout before the other. Other than that I agree.
For the first point you would need to make sure the balancer you use is throughput unlimited.
The second point is correct but it is extremely unlikely that one side runs out significantly faster than the other. If one side is out, the other is often not far behind@@AbsoluteHuman
This exactly
@@knicknacks That's just false, not all orepatches are symetrical, actually most of them simply aren't.
On the outside the ore patches almost always will have less space for miners.
With a belt balancer you can still use those sides without issue, without them you either need to create a complicated circuit network to somehow still balance the output.
Or just merge those outer belts into one of the other belts.
@teaser6089 You could slap on a single belt lane balancer on the belts that are significantly sideloaded before they go into the balancer. Or on all the belts before the balancer, for that matter, if it's a big issue on that particular orepatch.
9:36 See this is where your entire argument for not using balancers because of "variability" falls apart. The reason you are able to achieve perfect ratios is because the belts have a maximum throughput, which is defined by the type of belt. It doesn't matter how you get a full belt of ore, all that matters is that you have a constant full belt of ore. The only difference is how the system behaves when it starts to run dry, with no belt balancing, you are looking at some parts becoming variable before other parts, since they don't all have the same number of miners. Whereas, with belt balancing, the ones that have excess miners will pick up the slack for the lanes that are mined out, at least until you hit the critical point that you simply don't have enough ore being produced to oversaturate the system. But at that point, the mine is being mined out, which is always going to happen, regardless of your methods for extraction. Essentially, your argument is that belt balancers increase variability, when in reality they reduce it.
Also, 9:47 shows me why you don't like belt balancers, because you for some reason think that if you have 8 lanes being mined, you have to have 8 lanes being output. That will as you say induce variability very quickly. When I make a mine, I build as many lanes of miners as I can, then feed them into a N x 4 lane belt balancer. Which essentially gives the exact same result as what you are trying to achieve, except it's balancing across all the miners, to decrease flow variability in the system.
On the topic of trains introducing "variability", that is simply a matter of train network design. If you assign the proper amount of buffering, as well as the proper number of trains for the distance traversed, you should never have a lack of resources at the unload station, unless the load station isn't producing enough, which is just a matter of output not meeting input demand, belts have the same issue.
Ok, I have two questions. 1. How many different belt balancers do you generally use? 2. How do you achieve a perfectly even flow when the number of inputs and/or outputs is odd?
The train thing... I don't do buffering. The way I see it, if I have a train waiting for another train to be fully unloaded before it can do anything, why not get busy unloading it instead? Like instead of it waiting there buffering, construct a dedicated station for each and every train so that no train ever has to wait ever again. Wouldn't that be faster? That's the principle I use to design my rail networks. It is chaotic and it is highly variable.
@@theotherbigfoot So, I use prebuilt belt balancer blueprint books that I just slap into place, they have every combination of inputs / outputs you could possibly need. Using the blueprint books, if I have 7 lanes of ore being produced at a mine, I just slap down a 7x4 belt balancer blueprint. The people that made the blueprints did the work to allow every input lane to flow to every output lane. And that's the beauty of balancers, as long as you have 4 belts worth of input (for example 8 half full belts of input) then you can keep 4 output belts full.
On the topic of trains, I was mostly referring to having the right amount of buffer chests. But your comment has me thinking again that yes, there are actually two ways of buffering trains. If you prefer to not have multiple trains per station, that's perfectly doable, but from a variability standpoint, you are less likely to create variability if you oversaturate the system. It's the same principle behind you routing the mining belts through two times, it creates oversaturation, ensuring the belt is full for longer. Having extra trains running per station is the same concept, just with trains. But you can also just make the train longer, which equates to more buffer chests unloading at once, which leads to longer times between the station being empty, and longer allowance for the train to fill back up and get back to unload.
@@theotherbigfoot I personally just use one 4-4BP, 8-4 BP, and a 12-8 BP, since my trains are 4 wagons long.
As for buffering there's an extremely simple setup for train outputs: 6 blue inserters pulling out of the train car into 6 steel chests that have 6 more blue inserters putting them onto a belt. Eventually these become stack inserters if it's an extremely major station (for example the ore dropoff at your foundry). There are better setups for stations, but having this as your loading and dropoff is old reliable and does the job plenty well.
Having multiple trains on a line can very rarely be useful, too, but you should almost NEVER have a line long enough where the train hasn't brought another load in by the time the buffer chests are empty. Not until the end end endgame
Just... Don't have your supply run dry?
@@theotherbigfoot you don't need a train waiting for buffering you can buffer with chests
You seem to be overlooking a factor in that "designing for maximum flow" argument. (At least when ore trains come into play.)
You are correct in that *the act of balancing* won't solve the variability of the mines, only *compression* (i.e. overbuilding miners) will.
Which also means that if you belt directly from mine to factory, then you *can't* use an 8-to-8 or 4-to-4 balancer (to solve that problem). Because as soon as a single one of the input belts drops below a full belt of throughput, the output belts can never be full. The presence or absence of a balancer can't change that. In that situation you need a 12-to-8 or 10-to-8 or 8-to-4 balancer to guarantee that your belts are compressed (or more than 46 miners on _EVERY_ belt).
But the trains make the difference:
If you look at *one* outpost with an 8-to-8 balancer that loads into 8 wagons, then you can deduce that you'll almost never get the full 8 belt throughput going *into the train.* At first glance this _looks_ just as bad as directly belting ore *from miners into the factory* through an 8-to-8 balancer.
But even if the train doesn't *load* at it's full 8 belts, it can still *unload* at a full 8 belts. That's where the depletion of ore patches comes in to save us. Since players know that their ore patches will deplete, they tend to future proof by connecting more ore patches than they need at the moment. But that also means that more trains are loading (at under 8 belts throughput) than are unloading. In total, this _usually_ increases the rate at which ore can flow *into trains* to above the rate at which the dropoff stations take ore *from the trains.* (And _if_ the input rate is lower than the consumption rate, we call that "I need to add more mines".)
This overbuilding (more ore *patches* into the dropoff stations) provides the same compression as a 12-to-8 belt balancer in ore-patch-to-factory belts, or as if more than 46 miners are connected to one belt.
The main difference is that if you overbuild ore patches, and if you use balancers that compress down, like 12-to-8 balancers, then you will have a balancer between the ore patch and the train, but not with the "connect 56 miners to one belt" strategy. And with any balancer there, even an ore patch where only a single belt still carries ore can contribute to that input flow rate. Otherwise the train won't be able to fill all of its wagons.
I don't disagree with anything you say. All I'm doing is setting out my argument for why I don't use them. My argument isn't necessarily right, it's just my argument. I would really like to see someone put together a video titled 'why I do use belt balancers' because I think it would help me understand the issue better. I struggled to come up with reasonable use cases for this video; I don't use belt balancers, I've never used belt balancers, so trying to come up with good examples was nigh on impossible. I wouldn't even mind if it meant I had to remake this video, if there is something I have missed then I'd be happy to update it. But what I'm reading in all the comments is that everyone seems to have a different take on the subject. I'm not reading one argument, I'm reading many different ones. Some even agree with me in not using belt balancers but disagree with why. It seems to be a genuinely polarising discussion. Which I love by the way. Its awesome that not only do we all have different approaches but that we are also willing to spend a lot of time making comments on youtube that explain those different approaches.
The trains thing is something I want to cover in a dedicated trains video. It's relevant to this story, so I needed to mention it, but it isn't the subject I wanted to cover in detail. It isn't unreasonable to have expected me to cover it in more detail in this video, but hopefully once the trains video is done it will make a bit more sense. It might be a while though, I have at least 4 or 5 videos to do before then.
It's often a style choice, my dude. He likes building stuff a certain way.
@@Hawk7886 Everything is a style choice but that's not the point of the discussion.
@@theotherbigfootin my big bases I have trains for pretty much everything,
stations with multiple inputs sometimes multiple outputs,
I can't do this directly to train (on the scale that i like),
and i want to have more than 1 belt worth of inputs
so I use balancers (for input and output, though output is much more important, unbalanced trains are a pain :D)
Using a balancer with a large compression ratio (8:4 for instance) makes setting up that "guaranteed flow rate" much easier to automate. Instead of carefully adding a few miners to edge rows, balancing out the number of miners per belt, etc - just overbuild by ~1.5x and compress down to a balanced flow with a balancer. The balancer ensures that the central miners pick up the slack for the edge miners as they exhaust, and keeps a consistent flow going. You get the same result as your guaranteed flow rate miner layout - with a proper balancer and enough input lanes, the output will remain 100% compressed until the or patch is physically incapable of producing enough ore to meet demand.
Edit>
Say you have an ore patch that can sustain ~150 miners. You could manually group them into ~3 50 miner groups with some careful arrangement so all of them will run out at about the same pace. Or, you could group them into ~5 30 miner groups and compress down with a 5:3 balancer and get the exact same result - 3 output lanes which have extra "redundancy" so they will maintain their consistent output for an extended period of time.
For me just always use buffers everywhere, i know its not really what this video is about, but by buffering you cancel out those times where your patch "overproduces" so you just have to worry about "is my buffer full" if yes, the factory works on 100%. This also kills the uncertainty about trains, if you unload a train into some chests, the chests will always unload at the given speed while my train network can be chaotic
Imho the only place where buffers make sense are in train stations and energy production. Elsewhere the belts and assembler contents are more than enough buffer for most purposes.
@@roderik1990Agreed, overusing buffers can "hide" flow rate problems you have until it is too late and suddenly you're out of Iron plates
@roderik1990 if you use steam storage for nuclear, take a time to calculate heat capacity of all of the components that transfer that heat. I've got that 2x2 reactor with minimum (at least I really tried) h-piping and ceil ammount of heat exchangers in summary enough to store all of the heat that will be produced by one load of 4 fuel cells in each reactor and even slightly more than that. That fact elliminates the need for storing steam if you don't plan to use it as a battery for energy consumptions higher than reactor can produce.
@@roderik1990 yup you should do your manifolding at dedicated depots where resources are distributed. not at each individual input site where those resources are used. that's just wildly inefficient. you want your factories to have monoid like modularity.
Much like silos in a real factory. The incoming rate of raw material is variable, inconsistent. Building a reserve and pulling from it at predetermined rates allows downstream equipment to run efficiently.
This video is just... wrong. You don't use balancers to maximize output. Maybe at mining stations, but balancers are used at train unload stations because of the simple fact that it's virtually impossible to draw exactly evenly from the numerous belts coming out of your train. You are creating the inconsistency problem you are trying to avoid by refusing to use balancers.
Tbh, if you dont like balancers - use aai boxes and loaders. But its just my experience, and i dont find belts intriguing or interesting enough to mess with balancers
@@Neighterthere’s a belt balancer mod that just adds a little box that you put in your belts to balance them, it’s what I’ve always used
@@Megarah0 its a bit ups intensive from what ive read, but sure
@@Neighter someone released a “performance fixed” version and as far as I’ve experienced it’s not too bad
That said, the video isn't completely wrong, because you don't need balancers in mining outposts, flow routers (throughput-unlimited widgets) work just as well. Belt balancers are still quite useful after unloading trains though, but may not be necessary in rail city block designs.
What you are doing achieves what a balancer will achieve, only with more effort, thought, time, and variability. In your example ore patch, you are linking a specific set of miners on the outside of the patch to a specific smelting column. If you link that specific smelting column to a specific green circuit build, then, when those miners inevitably run dry before the center ones, your specific green circuit build will run dry. It's possible that the rest of the miners will be sitting idle, since the rest of your iron might depend on those green circuits to be flowing in order to be used.
If you used a balancer, then all miners will be supporting all furnaces. When some run dry, it won't matter, especially if you apply your "gut feeling" system to the entire ore patch. Your ore patch has 292 miners available as a maximum. This is 4.2 blue belts or 6.3 red belts. I tend to stick with red belts for a much longer time than I should, and you're using them in this video, so I'll use those as my example. Feeding these 8 belts containing a total throughput of 6.32 red belts into an 8-to-6 balancer will give you 6 fully compressed red belts for at least a little while. This is what you're doing, only this abstracts the whole patch into those 6 belts rather than connecting specific rows of miners to specific belts. Even better, use an 8-to-4 balancer and have a guaranteed 4 red belts of iron for a long time. That's what I do.
(Also, I included the 16 miners you very fairly labeled "annoying." You asked someone else how they'd handle an odd numbered belt. With this, I'd just lane balance it (your "scrambling widgets") and add it to all 8 belts one at a time, or combine it with the lane of 32 next to it, since all total that would be 31/s ore put on the 30/s belt, which is fine imo)
For me, balancers are all about the abstraction. I don't want specific production chains tied to specific miners. I would rather 1 production line using iron products draw from all iron sources equally. It's the difference between having everything slow down as a patch depletes vs having entire sections stop, potentially causing the entire base to stop. On top of that, I don't want to care that a specific row has 36 miners and comes out at a rate of 23.4/s. I just make sure there isn't a far bigger
(You don't seem to want to discuss trains, but I'm gonna add this in, since trains are the highest frequency use-case for balancers. Trains don't add "variability" if used properly. Using them the way you described does. But given a proper buffer and enough trains, there is no "travel time". A train should leave with items in the buffer and a new one should come in before the buffer depletes. This gives a very consistent output rate without the hassle of routing belts everywhere. It also allows for abstraction between ore patches, not just within.)
Wait but, your idea here as you are saying is to get rid of variability, but your method leads to the base having variability as the miners turn off at about the same rate that the balancers do if not maybe slightly faster since they cannot help for/account for each other. Though your way uses less resources in avoiding the balancer.
Also you can use a balancer that goes from higher inputs to lower outputs if you want to guarantee the consistency that you are looking for ^-^
Also if you use blue belts on the miners you could split it off into multiple red belts to really get the max mining at all times I just realized.
Interesting video. But this hasn't resolved the issue of depletion, it's only delayed how long it takes for you to feel the effects. You'll still have miners that dry out and slow down the production of random components in your factory. It's very interesting that you point out how trains introduce variability in flow rates, since that's precisely where most people are using belt balancers.
But I think you have the logic backwards. Trains are an additional source of variability, but they also are the best way to solve the main problem, which is miners that don't remain consistent. Trains allow you to dramatically overbuild mining outposts, which solves the problem at a different level. Use balancers to evenly extract from multiple mining outposts at the maximum rate locally, and overbuild mining outposts with train systems to ensure a consistent flow globally. Using maximum extraction rates locally guarantees that your trains remain balanced, while giving you the largest amount of buffer to keep flow rates stable at a global level.
The best part is, you can measure the entire system globally to ensure that you've got ample time to be alerted and set up new outposts before your excess capacity leads to down time inside the factory. Use a giant train stacker to simultaneously buffer full trains, and measure how many full trains you have using circuit networks. If the number of full trains starts dropping below some critical level because trains are waiting idle at outposts to pick up more ore, you know it's time to build more outposts.
With your system, there's no builtin way to realize that miners have depleted, without noticing low production at some part of your factory. You could measure the belts directly with circuit networks, but by the time you hear an alarm, it's already too late to add more miners to get 100% flow rate. Plus, I don't see how you handle ore patches that are so depleted that they can't supply a single belt. At some point you have to start combining the output of multiple mining patches. Trains are just a really convenient way to do that.
Trains solve the problem at a higher level, by taking advantage of their capacity to move far more material than belts. While you could still dramatically overbuild mining capacity without proper belt balancers, the miners will deplete highly unevenly because miners stay inactive when the belts are full, which leads to the same resource patch becoming irregular faster than if you're using balancers and trains to buffer the output. At the diagram at 10:00, the middle of the blue miners will take dramatically longer to mine up all the ore than all of the combined purple miners will, because the middle blue miners aren't going to start working until some of the other blue miners deplete, but they're sitting on the most sense patch of ore. If you built two similar mines and used balancers to evenly deplete the ore patch, and chests and trains to buffer the output, those two outposts would produce an even flow rate for much longer than two of your outposts would, because of uneven miner depletion.
Still, this is all pedantic. There's no "wrong" way to build a factory. People love watching DoshDoshington because he often does crazy and weird things that look stupid, because he's playing by the rules of some nonsense challenges. I enjoy watching his challenges because each one of his bases is usually much different, because he hates doing the same thing twice. People love your factories because they're all very regular, almost like mathematical art, and there's a sense of predictability to everything. Some people like large trains, some people like smaller trains. Some people don't even use trains at all, and prefer to use belts. It all comes down to personal preference.
As an aside, most people build train dropoff points so they supply an even belt of output. So if you know you're depleting all belts evenly, you don't need to use belt balancers on the output side, only on the input side where the miners are.
Then again, I've been playing around with a lot of different train unloading station designs, because it turns out that it's really hard to design a good unloading station, and most designs have some pretty serious flaws. It's non-trivial to build a train unloading station that can output a fully compressed belt, while evenly unloading all the chests. Many of the trivial circuit controls for even train unloading will fail if used in non-optimal conditions, or with different tech levels. That's likely why most people just build a balancer right after their train unloader, because it solves most of the design flaws with train unloading.
Plus, managing train networks can be scary, with the potential for deadlocks and all sorts of other chaotic problems. After playing around with trains for a long time, I definitely understand why some people prefer to use belts. To properly manage trains at scale requires some decent circuit knowledge and intimate knowledge of train routing. Which is likely why the new Factorio 2.0 update will have some pretty big improvements to trains and train automation, making the barrier to entry a lot lower to get into advanced train stuff. Trains are crazy complex for being such an early green science tech.
5:30 ...that's belt balancing. You're just doing it on the design side, which requires a very aesthetically unpleasant spaghetti of random belts connecting one or two machines here with one group waaaaay over there, rather than on the transportation side after you leave the mine. Imagine if you have two separate mines. Personally, I'd say "minimize the belts used to get the raw ore to the factory, then balance it so I can build a grid of equal size." If you decide to add belts to your design, you have to potentially entirely overhaul your belt network. If you balance, you just add another saturated belt.
I don't even play factorio - I play satisfactory, but the idea is the same. It's kind of like the question of manifolds vs load balancing. Factorio doesn't have that choice as much, but would you run a belt to each machine because it guarantees throughout? Nope. Manifolds are waaaay simpler to design, build, and expand.
As for all the stuff you're describing at 7:00, just reduce the post-balanced factory size by 20 or 30%. Or let it peak output early and slowly peter out, maybe saying "this section will be permanent but after the miners deplete, everything after this market will be starved so it can go"
It's balancing by other means than using a belt balancer.
this is where seablock is great. you build your factory with just the water, so no ore patches to add the variability
But isn't the role of buffers to even out the flow with variable inputs?
Train might deliver it's cargo in burst but the overall length of time it takes for a round trip is roughly the same. Thus you use buffers to accept all of it's cargo as fast as you can which you can then slowly distribute at even flow rate to the destination. Only issue is when your buffers become empty but all that means you need to increase input.
I'm skeptical of even finishing the video after the disaster of this failure in logic. Happy to see someone else pointed it out in the comments before I did. It's truly baffling how a person with the ability to play this game to such an advanced level can miss the idea of buffering. I'll hear out the argument about balancers but I do wonder if this guy could play more advanced mods with the stubborn attitude towards learning trains.
It's probably less of a concern in the very late game with very rich ore patches and high mining productivity, but no matter what you do, the ore patch _will_ deplete. You _will_ need to eventually build more mining outposts eventually and your existing patches won't be able to support the desired throughput. This is why I generally don't factor in ore rate into my calculations and just try to mine more than I need.
There's also the matter of an uneven _consumption_ rate. Ideally, consumption rate is much more easy to control than ore production rate, but it is still possible to have variable consumption rate. The most common example is in the mall. You generally don't consume resources making building that aren't being placed, and the buildings will only get made, and hence consume resources, when you're actively expanding the factory, or when it's building back up the buffer consumed in a recent expansion. There's also the possibility of uneven usage in the science path, but there using a balancer is little more than a way to hide the problem rather than an actual solution. The actual solution is to fix your ratios and builds.
I'll be honest here, I will get bored of a factory pretty quickly after I've finished building it. I want enough mining production without depletion to be able to test the SPM for a few hours but after that I'm done and I'm onto the next factory. I've done the outposts, I've sunk the time into doing all of that and I don't find it fun. I've never used belt balancers ever, so to be honest again, I don't really even know if what I'm saying here completely baulks the argument I make in the video.
The thing you say about variability in the shopping mall is absolutely true but I see the issue differently. I see my factories having two stages; 1. construction, and 2. SPM testing. I won't begin stage 2 until every pipe and every conveyor belt is completely full of ingredients. If ingredients are still being allotted to conveyor belts then that is taking resources away from science, as you say. If we are still extracting resources from the factory in order to build it then any test is pointless. We are still in the construction stage. So, yes, that variability exists for the factory, but it shouldn't exist at the moment of testing.
What kind of SPM do you shoot for in your factories may I ask? I play stock settings on vanilla and I have about 30 outposts in my 600SPM factory. I can't imagine relying on just the ore patches in proximity. Another way to play is to grow the factory over time. For me construction is ongoing so there is no construction stage and science stage. I started out around 120 SPM, built up to 240, and then went to 600. Next I might go to 1000SPM. I use trains for EVERYTHING and there is no variability in the flow rates into each section of the factory because the trains are over built and they buffer so the input belts are always 100% saturated. There are so many ways to play this game, like you said, and no right way. This is just my way for my current playthrough (on my 6th playthrough of the game, all the times before I used a main bus base but never got beyond maybe 120 SPM) @@theotherbigfoot
This reminds me of how it's preferable in starcraft to be undermining multiple bases at a time, rather than oversaturating a single base
Not when you're traumatised from losing fringe bases to run-bys. 🥲
this woud be a problem for me for like the first 1-2 hours of a playthrou. The moment i have Trains, 100% of my new ore patches are coming in by train and if i don't balance the belts bevore going into the buffer chest, they will forever be uneven and the train pickup time will be limited by those slower filling chests
Balance the chests logically instead of using splitters, works better. Also it will even itself out if one chest is more full than the others. Calculate the average content of every chest with one combinator and disable the inserters if their own chests content is larger than the average ammount.
While I understand your point, not using belt balancers at the volatile input of the system, be it trains or mining drills is kind of missing that point. Use a 7 to 4 balancer so you have a filled 4 belts for a long time and you can go from there with the 4 belts of ore output. Using them within the factory instead of just at the inputs is a crutch to have stuff going, even when throughput is insufficient, but that's okay, since things will eventually back up and it will naturally hit its maximum throughput. It's not the cleanest solution, but it is a solution nonetheless.
There is nothing to really understand in his point since its wrong... His max throughtput works because he did a 8:5 compression to a garanteed saturation. He could have achieved the same result with a 8:5 balancer without taking half a day to create clusters (which are also variable at some point, and which he will need to adjust every 2 hours anyway all around the map...)
i found that the best way to treat ore patch is to set to output north and south(east or west) so that the first ore to be mined is the center part. by the time the center most part is deplated, you will have consumed the most out of the patch so you basically obtain the max output for longer
Screw belts, just mine directly into provider chests and use manual labor to move heavy ore around xD
@@NaoyaYami_the screams of the bots intensify_
@@Hawk7886 Factorio 2: Rise of the Machines
Ballencers are like buffers, unneeded in most places, but critical in the few places they are needed.
Normally just with trains, and things that are basically trains, however some mods do add other things with odd pulses that this also applies to as well.
A ballencer is to make sure things load and unload evenly into buffers, that exist just to eat the variable flow of trains.
Best use is at a mine, where you set it up so a train will only show if there is enough to fill it, so its best to make sure each wagon has a buffer, and each buffer gets filled at the same rate. If you design your base correctly, you can prevent the need for balancers on the unloaders by designing around belts as the unit of consumption, but otherwise a balancer is useful to make sure the 3rd car isn't drained at half the rate.
Otherwise they are unnecessary, as the large numbers of trains, and well chosen points to balance, make them unneeded.
Sorry if I'm missing something obvious, but why not just set up an ore buffer near the smelting columns, and send out trains to mining outposts as needed to top off that ore buffer. The trains then just have a condition to stay until emptied.
Do the same thing at the outposts to minimize unnecessary downtime for the miners (if you have something larger than chests, maybe even enough to empty the patch), and you have a constant rate without carefully sectioning off your miners. Especially since they will start being depleted eventually anyway.
That's called 'factorio basics' and most amateurs have that figured out. *Baby's first standardization* is an inserter-chest-inserter setup going in and out of rail cars.
Personally I use balancers on my ore patches more to compress the input down. Sure the ore patch may have 8 lanes of mining drills, but some of those will have less output and will deplete sooner, so typically I will compress it down to 4 belts. If I need more than 4 belts I'll go find another ore patch; there's plenty of them! With a well-designed train system you can reasonably reliably know how long a train journey will take with only occasional traffic issues (and this'll be even easier with the expansion's elevated rails!) and use this to judge a suitable buffer size to compensate for the intermittent times. I very rarely use balancers outside of the mining outposts. Might use one or two in my main bus just to rebalance or compress belts down as resources are consumed (for example I'll compress my 8 copper belts down to 4 after the green circuit section) but otherwise I just rely on the fact that balanced belts in = balanced belts out! A full train unloading 4 belts of materials will unload the same amount of materials onto each belt, so there's no need to ever balance that. The output of that block will likewise be evenly split. Just... don't mess with it by hand-feeding some machines XD
balance is not about throughput. For example, if you have one line of green circuit, and want seperate flow for few items. With balancer you can make that, without wait each (one by one). It's not a big problem, but at start not too much resources (but there is a lot to be built).
Main use is trains. Most problem load/unload wagons/chests (>4) . In one chest zero items, others full, efficient lost. On this way need not only belt balancer, but belt-each-line balancer. But in vanilla it's not that necessary. Only for big_mods/megabases it gives you more time, before production will slow down, and you should personnaly check what/where happens. With overproduction it's still has no meaning, but only you design your base, and deside how it would work
I LOVE your furnace stack design, I think I'll try building them in my next playthrough
For me, i only use balancer to fill my ore train so when a train is stuck i know exactly that it is because the patch is totaly empty.Then i stock those train near my smelter zone in a full-of-ore-train buffer, to make sure travel time does not matter. And insiste my factory, no balancer, only perfect ratio as far as i can, nice video tanks a lot
I absolutely LOVE how the end message for orefield setups regarding "fixed, guaranteed flow" is literally "it's basically gut feel."
I totally get what you're saying. When I first played the game, I remember seeing posts regarding the maxim, "overproduce far upstream, it's way too stressful when you run out of source materials."
I've always found that true, for better or for worst. It can be super rewarding to play with absolute ratios 100% of the time, but I still have the most fun in factories where I suck up everything I can and work out the magic in the middle of the base.
You have point with the flow rate mindset, but only some belt balancer are "needed" because of variable input and/or output flow rates.
Most players use trains to transport ore into the base. The loading and unloading trains evenly is needed to ensure a constant flow rate. (or one cargo wagon will empty and the other are not)
Like you said, trains have variable "flow rate", but trains are a method for abstraction and modules in your factory.
Sure you can design a factory that create x science and calculate all the necessary inputs/miners/... , but it is much harder than building a factory and use some belts balancer and increase the input if you need it. Using belt balancer to much can be a indicater that someone missed the idea of ratios and can make the factory more complicated than needed.
You can't keep the same flow rate belting from a single ore patch, it will always dry up eventually if you play long enough. This is the beauty of trains for me, taking resources and distributing to where things need it, which acts as a sort of balancer. I used to rely a lot on balancers but they are kinda big and clunky, also things will balance themselves out if you let outputs fill up.
I just priority flow which ensures the whole belt gets used across the factory.
i really like your outlook. Everyone plays differently and you couldn't be more right. Like I use balancers everywhere, like an addiction. You use these precise rates but my method is nothing like it. I prefer to make a surplus of everything, if im not filling buffers, I'm not producing enough. Sure I need more power than a roided out strong man, but thats the way I like it. I end up wiht a similar result, just in more space. Youre factory looks incredible, your style is how a feel like a city block factory is meant to be made. Clean, efficient, and perfect. Love your perspective man
While balancing around maximum output has its problems I think in case of mining operations it is worth using balancers but with shrinking. I calculate the amount of drill output I get, deduct 20% of it and then create a balancer from however many lanes I've got to the 80% theoretical maximum throughput. This way in case of bigger patches they will last a long time, especially if more levels of mining productivity are researched.
belt balancers are eliminate variable flows. fun video!
edit: if you want a fixed flow rate for as long as possible from each mining outpost, the only solution would be to balance the miners so they all input onto a single yellow belt.
I might be a bit too new to the game, but this just does not make sense in any way. how does this balance your output, when this doesn't do basically anything? You get basically the same thing, but better with balancers. What am I missing here?
You're not missing anything, their argument makes no sense.
@@iantl2419 oh, alright
The problem with building extra miners per belt is that miners that aren't mining are wasting a small amount of electricity just by existing. When you put the ore in a train, it all equalizes out anyways, as long as you have a queue of trains coming in, your throughout is uniform.
In order to even out the output you can build a capacitor. First batch of ore is collected and put into chests. The belt gets full. When the belts start to dry out, the excess of ore collected at the beginning is now being put back on the belt
just crazy idea.
to keep a fixed flow for as long as possible, wouldn't it be better to try to prioritize the mining from the central patches since they have a higher richness than the outer ones?, so to try and mine the ore as evenly as possible to avoid the "outer edges running out" which would drop the mine output
Oh I actually like this idea
I generally don't use a balancer for maximum flow at a mining outpost. I usually use it to just spread 'load' across the patch, and to get the number of belts down to a number that makes sense for my train size.
❤❤❤ I think I love you! You are essentially cutting the maximum ore output from a patch in order to provide a longer period of consistent ore flow rate per belt. This means you might have 1 or 2 fewer output belts but your system is more predictable. It also means you are constraining your factory bottlenecks to the input and forcing the input to never exceed the maximum throughout that the system can support, and also buffering that input so that the system is less likely to be starved later on.
You know, one of the biggest problems I think I have is the mall idea because it encourages segregating materials for construction from materials for science and well as limiting the expansion and scalability of these items. Not to mention encouraging building along the main bus rather than formalising the idea of sub buses which are tailored to specific purposes.
I suspect I know the answer, but have you read "the goal"? I'd be fascinated to hear your impressions of that book.
If you design with factory-wide guaranteed flow you create these issues:
- every expansion requires full vertical implementation
- inability to prioritize production
- fragility - any belt going “slack” will ripple through all the production downstream, worst case scenario blocking entire production lines.
Trains also provide guaranteed flow. If one train can't being enough resources to the unload station to guarantee a constant flow, you can just add more trains to the same unload station.
This is a very interesting way of thinking. I like it because you can design a factory perfectly to maximize every resource gathered. Where this falls appart though is the moment you step into multiplayer 😂
What about equalizing both sides of the belt?
If you don't do that, one side of the drills might empty much faster than the other, leaving the remaining half of drills outputting down to half the original rate.
I don't want to be pedantic in return, and I don't want to say my friends are clowns, but like, we play together on a server and love it! But I think your argument isn't so much are they good, but are they precise
We don't sync up between plans so sometimes there's a miscommunication or misunderstanding or even just sometimes bad math because someone's baby is crying when the graphical calculators come out. Not to mention we play on different timezones so sometimes we're fundamentally in a place where things don't run as smoothly... Balancers allow us to just shore up supply, in a hodge podge spaghetti build that's gotten out of hand, despite the best of intentions, simply because there was an issue somewhere when we were planning weeks ago. Balancers alleviate the problem and while the impact is less production, we get more across the board than we would if one line was 100% fed and the rest aren't, which of course creates the need for balancers further down the road and the problem grows... But that's the nature of precision Vs good enough for some nerds being nerds to our hearts content in our spare time
I will accept that I did not consider the multiplayer angle. Fair.
Besides I'm not saying they are wrong, just why I don't use them, and I don't play multiplayer so it makes sense.
I am also pedantic and I like to be precise. Thinking about it, I don't think I would enjoy multiplayer.
@@theotherbigfoot I get you're not saying they are wrong, but that when building mega factories, we can do better, which is what you lay out and you're right. It just won't work for me and my friends and how being human gets in the way of us playing
Great video btw :)
Two mods that help a great deal: Miniloaders and Merging Chests. The second mod lets you slap down arrays of steel chests and then merge them into giant chests, warehouses and storage depots, and using miniloaders, it makes it easy to flow belts into and out of these large storage sites. The miniloaders pulling on the contents of these storage sites is done evenly. Thus an even output is obtained. I use these for my mining sites to give me a bunch of uniform, compressed belts of ore. The sites are also nice because they can double as in line buffers.
In the miner example why would you not just feed the 6 lines into a balancer and only pull 4 out this gives you a buffer and you can even setup logic to tell you when production is slowing down this will make sure the variability does not trickle down allowing you some time to setup another mine I personally don't use balancers, out side of miners, either I use belt compressors compressing the flow down to as few belts as possible and then splitting off the most compressed belt for production this make sure each section of the base gets full production and then adding more plate production if required
with the addition of the concept of buffers, it all depends on your expansion rate if you can keep up with your factory consumption rates. You could also just do on-site smelting to lighten the logistic load
Imo your clumping of miners is just a temporary solution. It guarantees full belts until the patch dwindles. Topping up such a belt is arduous.
I like to think of trains as that guarantee. I just need the sum of all my mining outposts production to be greater yhan my consumption. It doesn't matter if an outpost is dying and produces just 12 per second. Its about the total.
That's where balancers come in: loading and unloading trains evenly. If you don't guarantee an even unloading in particular, you may not have saturated belts. You will if you balance it, as the wagons are drained equally and thus the set amount of inserters is always all active.
I see no other way to guarantee saturated belts and guaranteed constant flow. Mines are by nature unable to provide that. Only once you use trains to bring it together you can guarantee constant full belts, no matter which mine produces how much at a given time.
Maybe i just don't understand what you're doing, but to me those are all temporary setups. Sure you double up on the miners, pushing the issue into the future, but it'll eventually creep up anyway. Am i missing something?
The simplest and best method that works for me is to have a balancer of input greater than the output, 6 - 4 for example. Then you just mix your outer most lanes into one to improve their longevity. By the time those 6 inputs can't feed your 4 outputs then the mine is getting pretty close to being drained. you might need to rejig the inputs a little but you're also at the point of getting ready to look for a new patch of ore if you haven't already.
always over supply the inputs to the required output.
My big train unloaders usually have 3-2 balancers from each side of the same train for compression.
Your argument is based on the idea that maximising the output of the mines (or other resources) into the factory (or its segments) directly will improve the output of your factory. This is an incorrect argument. The goal you want to achieve and the method of input are disconnected from each other. Balancing or not, you can achieve a fixed output, even introducing trains to the equation. The goal is to maintain a buffer so you have a fixed input into your factory. In doing so, you ensure that your mathematical perfection persists, because you know as long as you have a fully packed belt (or two, or four, or...) coming into your factory, you'll have a fully packed belt going out with whatever item you wanted.
Allow me to explain. But first:
- We'll ignore the complexity of the factory's internals itself, because that's a mathematical nature we don't have to address. We'll assume your ratios are correct. And with a compressed belt you'll achieve optimal output ratios for whatever you're intending to make, so ultimately that's the problem to solve.
- For the purpose of this counterpoint the definitions of "raw resources" and "resources" are synonymous as well. Copper wires are to green circuits as copper is to smelters as copper plates are to assemblers.
- Finally, a "construction unit" or just "unit" is any segment of your factory which consumes resources to produce something else. Doesn't matter if you have a unit that is exclusively copper ore which exports copper plates elsewhere or a self contained science unit which takes in all manner of raw resources and combines it on site into science for export.
---------
The assumption that you can ever feed a consistently full belt of resources into a unit directly from a source is wrong. Flow is always variable directly from a source, even if it's just a ripple effect from depleting resources in some fashion or another earlier in the chain or an overabundance of resources clogging up your current unit because it mistakenly assumes it can clear enough materials to make and deal with byproducts (i.e. ye olde oil processing and dealing with multiple oil types and petroleum). Maybe you add another unit that consumes more of a particular type of resource (like another thing that requires green circuits but don't get around to making more green circuits to accommodate). Maybe you don't notice the gradually falling off oil output from a field getting to its minimum and things progress as you expect until they don't and bottlenecks form that require you to start tracing backward.
The goal is to maintain the full and complete packing of a belt (or series of belts) heading into a unit. That is all. To properly achieve this, you must use a buffer, and that buffer must (net) fill faster than the unit consumes it. Regardless of the potential burst output of a miner array or unit, as long as the input into another unit is always full, you've achieve YOUR stated goal.
This is where direct feeding versus belt balancing is up for debate. I'm in favour of belt balancing because it ensures a number of conditions are met:
1. When input is substantial enough, there will never be a blockage because the balancer ensures that surplus goes to alternate paths.
2. When input is insufficient, you will still get output from your unit, starting with a gradual and equal reduction across your unit down to a slow trickle from early portions of smelters or assemblers, ensuring every belt has functional components at least.
3. It also ensures that no one source will inhibit the input or output of your unit. For example: You have 4 belts leading into parts of a unit. For inexplicable reasons, one path consumes resources faster than another (maybe you miscounted a number of smelters, maybe you partially upgraded belts/inserters on one path but not the others, etc.). If the line going into that path wanes, it will produce less resources than all the other paths overall, especially when it consumes all the resources it needs on its path and becomes deadlocked while every other path chugs on merrily.
Belt balancers aren't always necessary though. What's important is making sure that the inputs to buffers are properly balanced. If you can do that without a balancer, power to you. If you DO use balancers though you can ensure that trains are filled equally and optimally in the shortest period of time. If you think of trains as a type of buffer, then you want to ensure that each car is filled as fast as possible to get that buffer to your units as fast as possible. Trains are "burst buffers", large-ish packets of resources which are meant to be offloaded quickly into local chests (another buffer) and then swapped for another train. The idea is you have an excess of trains supplying the resources your unit needs to function in more than sufficient quantity so that when the time comes that a ripple occurs, your factory doesn't notice it and it stops at the level where the ripple originated. At SOME point, you need to balance your resources to ensure an equal distribution of resources for your units. If your balance is just stuffing a train full of resources and then pulling it off of the train at its destination into chests, that's balance enough.
tl;dr: Balancers aren't important, maintaining a buffer is. Balancers can ensure an equal distribution of resources across your buffers. But merely stuffing as much as you can into the buffer in as short a time as possible and ensuring the buffer does not deplete is enough to ensure full compression. Balancers beyond your buffers just ensure optimal use of factory construction units regardless of whether a segment of your buffer is depleted early or not.
in the begning i felt lost, and at the end i felt like i was just starting.
That chaos added by trains is actually good! Your ptch started to run dry? Add another patch with the same name! This way both of them whould overproduce and only send trains when they are full, they provide constant flow of resources
There is always a bottleneck, for me that bottleneck is the belt, and I have no problem with putting too much on a belt and letting that balance things out.
A lot of really long comments in the comment section :D here's my long comment:
I use belt balancers often, but only for loading and unloading train cargo. Like you said, "trains introduce variability to the flow", and one way you can reduce that variability is to place down chests next to train stations that act as "buffers". Items are placed into the buffer chests while the train is away, and when the train is present, stack/bulk inserters quickly transfer items from the buffer chests to the cargo wagon. If you want to maximize the speed at which the bulk inserters transfer the items from the buffer chests to the cargo wagon, the items need to be evenly distributed in the buffer chests. And the only way to evenly distribute these items evenly is with belt balancers.
what you need is intentional grouping of mines, which you seem to do already based on "feel" rather than analysis and intent...
and a compression-balancer or reduction-balancer....something that takes multiple FULL lines and compresses down to a fewer number of lines (while also balancing lanes) such that, as the miners deplete, the output remains compressed. This becomes easier as mining productivity increases...and once it's to level 15+ becomes almost meaningless/pointless as the mining prod makes each line require fewer and fewer miners to compress a full belt....proper grouping and mixing/lane balancing still gets the job done.
Also...use of trains at stations and proper balancing and buffer design/unload mixing will ensure your base has a max flow rate......if properly designed... and if you stay ahead of the rate that you deplete outposts. Part of the problem you face is you aren't building ore outposts frequently/soon enough to mitigate the effect of depleting ore fields.
My current strategy: If an ore patch has a maximum throughput of X number of belts, I'll throughput limit that patch to between 20-30%(rounded up or down to the nearest integer greater than zero) by compressing the belts. I like the slow burn.
This is a perfect example of ovethinking things lol
Diagnosis: Correct. I'm an engineer, I overthink stuff for a living.
I think this is very odd. You can get away with not using balancers inside the factory just by balancing your train cargo and then in the dropoff you know you'll have n lanes of fully saturated belts. You don't need to do weird stuff with mining patches to ensure saturated belts
You are a very unique person indeed, i like your way of thinking
Not that i agree, but i like it
I’ve never felt the need to belt balance, but I would only ever need it for ore… and I just smelt ore on site and train it back to base to maximize efficiency. I also move my base closer to the resources as the game goes on since distance is the largest killer of throughput. I tend to do a lot of buffering and using “belt wrappers” instead of balancers. I plan for belts to always haul maximum items and when they aren’t, I release buffers first or belt wrap a new source of the item. But, this maximized line is also why I tend to avoid overuse of “blocks” and “main bus”. Each line is treated as a full production line. Spare items that won’t be used on a full production line are rolled back into buffers or “belt wrappers” to fill back in on lines where production has dropped.
But, I love the look of full belts even if they aren’t moving. I don’t like “balancing” the belt. I like it full.
Thank you for reasoning this through! I hadn't realised this was the solution to the problem i was having with flow rates - i.e. I was building my factory based on the original maximum rate. I really like your solution and will impliment it in my next factory for waaaay better control. Awesome man, thank you! :)
I have always struggled with this topic in Factorio. I'm definitely unreasonably pedantic myself, and belt balancing is one of the biggest road blocks I've had in Factorio.
Definitely a sensitive topic though, hope you won't be getting too much 'feedback'.
I hope I can give this challenge a fresh start with Space Age after watching some of your videos. We seem to think in very similar ways.
Thanks for your content!
I just found your video and this is very interesting (and contrary to those vids that I've watched before, bo other content creators here), I am going to trust my gut and give you, my good man, a sub, and carry on watching.
3rd comment i know but holy crap i just realized you are building your factories exactly like i am! brace for long comment. i sought to modularize my factory by blueprinting linearly mapped (16x16, 32x32, 64x64, 128x128) 'grids' of production lines separated and connected in a repeating grid pattern bordered by rail lines. i treat these individual grids like monoids. meaning each grid does not depend on other grids existing in any certain order or priority or placement thus "associative" if you will allow the monoid analogy. then i can just haphazardly place these production line grids in any sequence that i wish. a depot is included where outputs are gathered by train, buffered or balanced, and sent out by train. i designed the lines such that trains can never be path blocked. there is always a path because each grid connection is the rail line itself in a repeating pattern. no proof but i intuit this is the case. whereas most players build fragile rube goldberg devices. a monoidal factory is the most robust factory one can build. drop a nuke on any random spot and the rest of the factory will still keep going. i stopped playing the game because i am obsessed with efficiency and robustness. so i sought to optimize grid production line space by figuring out the best configuration for parts in the grid that would pack the most information. so i set out to write a genetic algorithm to do it but my lack of technical understanding of how to optimize this packing problem has stalled me and ive been putting it off for years. thanks for jogging my memory!
The blue belt has 8 item room in one square, the steel chest has 48x50 or 100 item in stack. The blue belt can move 45 item/s to the next square. The green arm can round 864°/s, so it can round 360° in 0.4s and it moves 10 item to 2 square distance, so the speed is 2 square distance/0.4s*10 item = 48item/s to the next square.
I like this a lot, i think its basically what ive always done just smarter and more intentional. I don't quite understand the drill blocks yet but i think ill try it in that manner!
The only place I use balancers is for ore. I just make lines of drills and however many lines I end up with I just stamp down a balancer that compresses it down to the target amount of belts. Say I want 4 belts out, but I have 8 belts of ore, I just make an 8 to 4 balancer. In this way, the only time you see a reduced output is if the total operational miners aren’t sufficient.
"My priority is mathematical precision"
"This is not an accurate system by any means"
My guy...
Are you suggesting that using belt balancers WILL provide mathematical precision? Because I have a problem with that statement the same way you probably have a problem with mine.
As a beginner, I use balancers. I don't even know what exactly awaits in the late game. I don't have later item recipes unlocked and my planning ahead is based a lot on peoples recommendations. Started to build a main bus base after spending like 30 hours on the tutorial and the started base for free play. I just have 60 copper per second smelter and also 60 iron puss another 60 goes into steel. The mining is just slight overkill, so I could run the furnaces.
I have no Idea how big I should build my next productions really. I don't know how much materials the later things will need. But it's fun to discover. I left some room for smelter expansion. But the idea behind the build was that someone in a video said that this amount is good for completing the game.
But after I learn how the game works, I like the idea of more precision and neatness.
I find this argument really interesting. I tend to use balancers on mines, and then use trains from there to the factory, but from the unloading, I don't use balancers ever. It's all flow rate, and one belt of copper here, two belts of iron there, and so on. I guess the trains make it so the variability of the mines is kindof a moot point, whereas inside the factory, precision governs all. Thank you for reminding me!
this is a nice philosophy regarding ore throughput
Discrete values aren’t just a miracle. The main juice of trains that they can guarantee constant value every time they arrive. Problems like “there is not enough throughput in this station” lies in the field of “Build moar trains. Build moar miners. Build better upload/unload designs”.
I agree, that in one production block u can (and i’d say) must build without any belt balancing (cuz all balance should be put into mathematics of ratios).
But when I need to make my trains leave mining outpost every 20-30 seconds with 4-inf train carts full - i need to make all buffer chests sufficiently equipped in approximately same time.
Belt balancers (or belt splitters in that case) are used.
Bro basically sais: I dont want to balance output of whole patch of resources, I wanna balance it by dividing into some balanced pieces. Plus I don’t care about imperfections of ore distribution through the patch. Plus I wouldn’t think about how depletion of even one drill will affect whole scheme.
“It’s most of the gut-feeling.” + “I want mathematical guarantees “. Schizophrenia.
Very interesting video with cool concepts. I always make more machines or miners that would over-saturate the belt. I always plan to make more than the output is capable of handling, this way I can be sure that, if I supply more than enough resources, I'll always get a belt's worth of products. I think my approach is very similar to yours, and it's something that really works. But the biggest issue is when consuming the belts, some can get less saturated than the other ones due to side-loading, imperfect ratios on recipes and other weird factorio cases. I don't use a lot of multi-belt balancers but I do use a lot of side-balancers
I appreciate your gameplay approach but notice some practical issues, especially with the fluctuating needs of your factory. Constantly monitoring each conveyor belt for specific processes, especially with raw materials, becomes impractical. My own experience led me from striving for utmost optimization to realizing that a simpler approach was more time-efficient.
When playing in a team, it's unrealistic to expect everyone to adopt the same building style or to understand the intended use of each conveyor belt. While communication is vital, setting basic standards for things like directionality, the use of main buses, and robots is essential. Overly strict rules, like those regarding belt balancers, can backfire. Although your strategy may work well for solo play, it might not be as effective in a team setting.
My gameplay has evolved to prioritize practicality, focusing on being quick and efficient, yet simple. This shift is similar to the reasons behind Python's popularity: its straightforwardness and readability, despite its performance limitations due to being an interpreted language.
okay, sure, you want a guaranteed flow in the factory, but as you pointed out the issue is at the mines because of uneven ore patches, which as you continue to play new ones get built farther from the factory. So use trains, and use belt balancers at the mines to evenly distribute ore into (the buffers that load into) the train cars. So it really doesn't matter what the maximum or optimal flow possible from extracted ore, what matters is that every train car gets loaded as efficiently as possible, which is why even distribution is what you want.
that little stint from 7:15 and on had was hilarious more of that please.
also, because inserters move resources faster between inventories (cargo wagons and chests) then they can pull or insert onto belts...the use of buffer chests at ore patches and at unloading stations....and proper splitting/balancing at BOTH ends is required to maximize compression on all belts in your factory. Skipping this is one reason why you may be forced into the mindset you've adopted.
I see your error. You said you dislike trains, and that's the same reason as balancers. You want every factory to run full output at all times, rather than calculate for max potential. *Buffer chests* are your best friend for that purpose. You say "I can connect one belt" well the rest of us can make 4 carloads and ship it to the other side of the world, and have a consistently full belt at the other end.
You also seem to misunderstand what balancers are even for to begin wih. The MOST important use of balancers, is splitting ore belts into trains. So that every car loads X stacks, instead of 7 stacks in one and 3 in another.
Or going into smelting, you don't even WANT 8-8, 6-6, 12-12, etc. you want 8-4, 6-8, etc. splitting input equally between production facilities.
As an experienced factorio player, this is lot of blabla just to say: I love doing a lot more work than necessary
Usual process: slap a lot of mining drills on a patch, balance it into chests, load into train, unload, done, now you can focus on better things than micromanaging drills
Oh and a patch runs out? Add another patch to the train network :)
a side by side comparison of 2 small factories running from similar sized ore patches one using balancing and one not
that would be interesting
I have a similar mindset myself, tho I use trains so the reason is different (I want train to be loaded as fast and equally as possible so they enter and leave stations as possible). But I goes on to saturate my belts from ore patch differently: I use priority splitters, in the example case I would also take 4 (maybe 3) belt out of the patch, and just load as much as possible from the 6 row of miners into 4 lanes, it's easier to setup and make identifying dying outpost quicker (one belt is obviously empty).
trains have higher bandwidth but also introduce latency. ideally when a train leaves dropoff station because it's empty it would be immediately replaced by a new full one that is right behind it, this plus the chests as a buffer should eliminate variability. on the mining side, build more mines than you need, so each mine works at full speed for half the time and then waits for a train. then as their output declines other mines will pick up the slack. basically just make sure mines have the ability to mine more than you need and then bottleneck them, if not build more ahead of time. I haven't played in over a year btw, so maybe this only works with circuit networks that set number of trains per station based on how much ore is in the chests. hope this makes sense.
Honestly I don't agree on the hate that being showed here in the comments. I found the video very entertaining and I like your logic.
I'm in no way a Factorio veteran, but I like the game as a casual, and I found myself always using belt baencers everywhere in my ore patches and every time there was a problem within a few hours due to less output, the very problem you pointed out in your video.
I think the next time I'm going to play, I'm gonna think more long term and try to do something different, like you did.
Great video!
That is a cool way to think about things. Great video!
what i personally like to do, is to use trains from ore patches, to a central smelting station, and hooking up speakers to the ore loaders, so that when the miners can't keep up with the trains, aka more ore is being smelted than mined, i have time to make a new mining outpost. Usually i have about 3 mines for iron and copper at any given time, and with this method i constantly have a full smelting setup, and reliable plate output. IMO your method requires way too much thinking, for a problem that could be solved easier, however, nothing wrong with either method.
And yes to the guy that's about to post "plates stack 2x ores, so moving ores in trains are ineffecient", go do your own thing:)
Seems like a hassle. I just use two blueprints to mine ore, the mining half and the train half. The first just throws down a massive square of all the miners that will fit, power, roboports, beacons, and an excess of belts. I literally don't care if massive lengths of belts/beacons/power connect to nothing, it isn't worth the time to trim it once your factory is big enough to crap out construction materials faster than you can spend them. Then I have a second blueprint containing a train stop (pre-named as an ore type), train queue, buffering and loading on both sides of the track, a balancer, power, roboports, and a radar. I can just shove down multiples of the mining blueprint and a train blueprint (pick the right one to avoid having to re-name the station) and most of the work is done. I even have a generic mixed ore variant for the overlapping patches. Then just lazily connect the belts between each blueprint. Really, the hardest part is making sure that really remote mining outposts have separate logistics networks supplied by trains to prevent bots from getting stuck. It doesn't have to be some perfect ratio, because it's fast and easy to just build more mining outposts. You still want the balancer to make sure trains are loaded evenly and not sitting there waiting for miners to work on the last car, or never finishing loading it at all. I've sometimes added circuit conditions and alarms to notify me when the patch finally dries up so I can pull most of it back up again at my leisure. You can have the train stop disabled when the buffers aren't full enough to fill a train, but alternatively you can just have a network with plenty of trains and train queues at both ends. Far into repeated research this setup gets replaced by direct to train mining, but I usually play for mod content and not repeating research.
tldr: You can't perfectly balance everything in Factorio, and if you try, my unbalanced factory will be bigger given the same amount of time anyways. Just quickly blast biters using spidertrons/artillery, expand/build more networks (power, tracks, logistics) and defenses, and quickly slap down extra mining outposts and ore trains. Have an excess of early production and taper off towards end products and science output. The alternative is factories stalling due to missing ore, which we would call a skill issue.
Initial setup: 8 to 8 lanes balancer
"no balancer" setup: 5 lanes
hear me out...
8 to 5 balancer. Guaranteed flowrate AND balanced mining through all fields.
2 things really. 1. Is a 8x5 balancer really possible? A splitter has two outputs so in my mind a belt balancer is only really possible in multiples of two. 8x6 and 8x4 I understand, but not 8x5. Can one be jerry rigged? Yes. Is it possible to build a perfect 8x5 balancer? Not in my opinion.
2. How many belt balancers is one expected to lug around? There are a lot of different configurations of belt balancers out there, either I remember them all by heart or I have a blueprint book filled with dozens of the things. I resent having to carry around all my perimeter wall segments and there's only 8 of those.
It definitely feels to me like belt balancers are one of those things where if you do use them then you will find opportunities everywhere, but if you don't, then you'll find good reasons not to. There are compelling cases for both sides. Nobody is wrong but everybody is stubborn!!
@@theotherbigfoot1. Google is your friend, 2. I don't see the point, blueprint books can be stashed in the blueprint library and then you don't have to worry about it,
I think balancers are a great tool especially with late game factories that use busses with multiple lanes of a single resource to satisfy the high demand that infinite research requires, it means you don't have to terminate lanes or use undergrounds to access most of your throughput
You can just use buffers to have flow garantees on your train outputs.
You can have guaranteed flow rates by invreasing your extraction when you are running low too...
No need to complicate things that much for guaranteeing flow rates.
I use belt balancers simply to avoid having to spend extra time planning around ore patches, which by their very design, are random. Even if I only save 5 minutes at each extraction facility, that adds up to hours worth of time saved pretty quickly. For me, I care more about the time efficacy that belt balancers bring.
You cant mention mathematical precision at 3:30 and the go with gut feeling at 7:18.
In real world engineering I frequently abandon the numbers and go with my gut instinct. It isn't as unreasonable as you might think.
dude i stopped at that moment too. hilarious
If you don' like variable flows, another solution would be to use a cascade of splitters with priorities always to the same side, so you get a bunch of belts with guaranteed flow, plus a few more belts "extra" just to fill the gaps. This way you don't need to "balance" the amount of miners for each belt ;-)
Interesting question, I personally will stick with balancers and buffer but was good food for thought.
Cool take, mate. I can't be bothered with perfect rations in my factories, especially when playing modded, but I often use something like 8 to 4 balancers for train loading, limiting the thoughput, but making sure that the particular mine will have constant output for a long period of time. I feel that's kind of what you are trying to achieve, but balancers reducing the number of belts seem better to me than combining mines manually.
This isn’t to say that your method is wrong, it isn’t but belt balancers can be blueprinted and provide an easy solution to a problem that is fundamentally unbalanced to begin with. In order to use your method we would have to calculate how many drills we need and add 20 percent on top to account for losses over time. And as the ore patch depletes the amount of ore will run out anyways. The only time this seems like it could apply would be if you go absurdly far away from spawn where the ore patches are all but guaranteed to run out. Only then would I see a practical reason to do this. For spawn or nearby spawn it would just make more sense to use a belt balancer
I get what your saying, but through the use of balancers you can achieve a guaranteed throughput. You do have a unique way of connecting your miners which makes more sense I think.
You can get the configuration Like that that you have for example 8 lines coming Out and you Balance Them to 4 so If one Line Runs Out you still will have 4 full Lines
Overbuild and multiplex the supply that enters the start of the process.
Been playing Sea Block and that's so interconnected with feedforward and feedback loops that no real attempt at math can work for me. In SB the consumption creates needed byproducts used in supply as well as side products leaving the only viable distribution method to be oversupply. I'm not even close to worrying about belt saturation or balance.
My playstyle is "need more, build more" and I'm sure that drives veterans nuts when they encounter it lol.
I think that is the primary philosophy of most veterans
I tend to just push the ore as far to one side of the set of belts as possible. build each furnace stack for exactly one belt. doesn't cause idling miners on mining prod. bonuses, and isn't too hard to patch new input/outputs, and will pretty gracefully handle any in/out mismatch. It is pretty splitter-heavy, but if I'm dealing with enough ore that it's a concern, I have enough ore that it isn't a concern. Scales pretty nicely for high science cost multiplier
I'm pretty sure that if each of us made this exact same video, whether for belt balancers or against them, each of those videos would be completely different even if the videos were meant to agree with each other. It's one of those issues that has an enormous variety of opinions.
This is just my take on the subject. It's only really relevant because I need to refer to it in another video I am working on.
@@theotherbigfoot the insane variety of valid approaches is one of my favourite things about Factorio! I love learning the ways other people approach the game, and is one of the things that really makes me enjoy (and occasionally absolutely despise) multiplayer :3
@@ShadowKestrel Agreed 100%
I do the same for busses, always push everything to one side so that when variability does happen you know how it will happen. But I still use belt balancers in other places, especially in train loading/unloading stations, because I often want to minimize the amount of time a station is occupied by any given train. And lane balancers (scrambling widgets) are important in lots of places.
I like this view.. some of these tips are actually more chill.
but at some point the ore patch will still be running out and ruin your perfect ratios
No system is perfect. Each will have its advantages and disadvantages.
@@theotherbigfoot I agree on that :)
The good thing is that we can play Factorio however we want!
Did you play Seablock? Since you have to create the ores and not mine them you'd have full control over virtually everything in your factory.
@@randomguy-no5vh Absolutely agree. I've been reading through the comments as they arrive, everyone has a completely different take on the subject. Its awesome.
I briefly looked at seablock a year or so ago but for whatever reason I didn't pursue it. It's not on my list I'm afraid. The approx plan is K2SE (which I've just started), DLC and then py. That should cover the next year or so.
I tend to slap down miners so that the entire patch is covered and just use a widget to get it it down to 4 belts that then toss into boxes. There are only 2 goals. 1) that the entire patch will eventually be mined and 2) that the train can stop and get full cargo quickly due to plenty of ore waiting to be loaded.. I do like the green circuit thing but seeing copper wire on a belt gives me the willies. Im a direct insert guy as much as I can possibly do it. In order are Direct Insert > Belts > Trains >>>>> Bots. I find the bots boring.
I'd rather just consume a belt or lane and shift the remaining lanes over. Better compression, and the builds are already ordered by requirements, so it won't affect output.
I think the idea is to overbuild the mines, make a balancer that fit your "fixed ratio" design, and basically not worry about it until the mine is depleting at that point you should already be feeding it from different source (usually trains) or the ratio go down, the fastest/close-to-automation way to build a mine and forget about it, even with your design the mines will eventually go down and the ratio go down. only when you have less mines to saturate the belts balancers would be unnecessary.
And for the second half with the input/output blocks (city blocks without trains) this is what happen when you move away of the main-buss-design public meta. which is a balancers hive. I agree on play however you find it fun. And I find your factory blocks fun to look at. they look like beautiful Art.
personal though, Unless you are playing in marathon, Going mega base or playing in a mod overhaul that push you to build a mega base, most those design patterns are just overkill/just-for-fun/Art.
Why build a factory based on its input and not its output? My approach would be to achieve a specific amount of output produced and to ensure that there is an excess amount of input to feed it. Excess input only really comes at the cost of power, and even then not always as things will stop when maximum throughput is achieved. In the case of trains, I use buffered storage to smooth out the variability and ensure that there is enough throughput to maintain a consistent flow of resources to the factory.