Fluid Mechanics Still Broken, 6 Years Later?

[Tertius]

Yeah, I am not talking about the OP, and I get why buffer tanks should be used. I talk about the tank's load not being balanced, ie. connect your wagons with a pump each and one taking alot longer to clear/fill: https://youtu.be/zJBvw28bQu0?t=1721

But this behavior is not a flaw of the game. It's not evil. Remember, the game is a (simplified) simulation of the real world physics and mechanics. If you take the example setup from the video to the real world, build similar tanks and pumps and controls, and inspect pump and fluid behavior, you will see the real world will behave the same as observed in the game. So this is a genuine game logistics challenge to solve, not some bug or flaw to overcome.

(the OP is about something different, by the way. It's about some detail that is not simulated correctly by the game and slightly different in the real world)

[FuryoftheStars]

[...]i'm not sure if there was priority on splitter back then[...]

There wasn't, but you were able to use a single circuit wire connected between the two output belts to create something similar. One belt would always be on, reading on hold, while the other will enable only when the signal exceeds the desired amount (so you didn't necessarily need to wait for the other side to be completely saturated before outputting to the overflow side, if you wanted).

And technically, of course, you can still do this.

[mmmPI]

There wasn't, but you were able to use a single circuit wire connected between the two output belts to create something similar. One belt would always be on, reading on hold, while the other will enable only when the signal exceeds the desired amount (so you didn't necessarily need to wait for the other side to be completely saturated before outputting to the overflow side, if you wanted).

And technically, of course, you can still do this.

That's how i would do it now, with barrels =) but back then, i couldn't use circuits, i learned thanks to the oil processing That's a neat trick. I'm more used to put everything on 1 side of the belt, and side load on this side, "if there is some room" to make the side-loading belt to "enable" only if the main is empty.

That could be replicated with a top-up valve, from mods, or a main pipe lane with a tank and another pipe lane feeding into this tank, but with a pump, only enabled when the tank is under a threshold or empty without mods otherwise disabled so as to isolate the other side loading pipe lane

[MeduSalem]

Meh. This topic is still going on after all this time? ^^


But yea, as others wrote already... just make more fluid than is drained; then it will hardly be an issue.


That said I don't know if they ever made the change to make each pipe-system act like one single big storage tank which providers input to and consumers take from. I would have liked that.
But I remember something like the flow direction for graphics being an issue then and that is why they didn't implement it that way.
Though I would have solved the flow direction issue by using a cached directed graph that has no other purpose than to fake the direction for the flow graphics and which I wouldn't give a damn about if it wasn't physically correct all the time. ^^

[coppercoil]

But yea, as others wrote already... just make more fluid than is drained; then it will hardly be an issue.

No, there's no such state in the game "I have enough of ..." . If your factory expands, you always, always lack of some components. "Just produce more of" is a temporary solution that will move the bottleneck to other place, and past some time it will come back again. Very often you cannot start producing more right now, so you have to distribute limited resources, this is the constant state of the game. Distributing liquids is counterintuitive, so it's not a just simplified simulation of the world physics, this is a bad simplification of the physics.
Yes, I can live with that. I learned how fluids behave and I know how to solve things. That doesn't mean fluids do not need fixing.

[Tertius]

No, there's no such state in the game "I have enough of ..." . If your factory expands, you always, always lack of some components.

With thorough and sufficient planning, you know the exact input and output of your factory, so you can provide this and extract exactly this.
While factory growing in the earlier game phases this is not feasible, but if you go for bigger and permanent buildings, you're able to plan this.

Usually, there is the problem that you don't know how much you need of some item, because you don't know much the goal requires in the end. If your goal is to launch a rocket, how many ore mines do you need to provide enough ore for smelting, for building the factory, for research, for creating the rocket and satellite?
There are numbers in the wiki, for example https://wiki.factorio.com/Rocket_part, but these numbers have no meaning if you're not able to convert this into amount of assembling machines, amount of mines, amount of refineries etc.

So you start building some estimate, and you usually estimate too small.

To get a different view and for transparency, I encourage you to go into map editor mode and do a factory upside down on a green table. In reverse. Start with the rocket silo. Provide the input items with infinity chests. Once your rocket silo works, continue to build production lines to provide the products you currently pull from infinity chests. And provide their input items from infinity chests. And so on. This way you always directly see how many input items you actually need, so you can design the production lines exactly for this amount.

In the end, you will only have infinity chests left that provide ore. Then add remote stations that collect ore (from infinity chests) and deliver this via train to your base instead of providing ore directly. Then finally replace the infinity chests in the remote stations with miners that collect ore from resource patches.
This finished factory will give you a feeling how much of precursor items is required to produce the next level of refined items.

The result of that green table build can be a bunch of self developed blueprints for production lines that all have a well defined input and output, and that match if you connect them. Might help you then to build a real factory on a real map.

[coppercoil]

I think most players don't play in the way you described. They play not "until the rocket" or "until XXX SPM", but "I expand the factory until I start new game". There's no concrete final plan for many players.
But let's assume the player plays just to launch the rocket. That doesn't mean that player known what are quantities. That doesn't mean that player wants to known them, he just plays and solves issues as they arise. This is very nice way to play the Factorio.

[FuryoftheStars]

But yea, as others wrote already... just make more fluid than is drained; then it will hardly be an issue.

No, there's no such state in the game "I have enough of ..." . If your factory expands, you always, always lack of some components. "Just produce more of" is a temporary solution that will move the bottleneck to other place, and past some time it will come back again. Very often you cannot start producing more right now, so you have to distribute limited resources, this is the constant state of the game. Distributing liquids is counterintuitive, so it's not a just simplified simulation of the world physics, this is a bad simplification of the physics.
Yes, I can live with that. I learned how fluids behave and I know how to solve things. That doesn't mean fluids do not need fixing.

1. As you progress in your factory building, you should start getting a fairly good idea of how much you're consuming vs producing. So, unless you just build and expand with reckless abandon, you should know when you're getting close to consuming everything that you can produce and should actually start working on finding more to produce first, as otherwise everything you build to consume will underperform and be something of a waste.
2. And even if you do end up consuming more than you can produce, the odd mechanics won't actually hurt anything (typically). You'll just end up with some of the lines you setup consuming all of it while some are left idle vs everything running at under 100% (ie, 4 at 100% and 1 at 0% vs 5 at 80%). Is it ideal? No, but you'll still be consuming and producing the same amount in the end.
   1. If the lines that aren't getting any of the resources are producing something that the rest aren't, unless it's for something that doesn't eventually get consumed into the full production chain (ie, military defense), then the rest of the lines will eventually backup as they overproduce and go to your lines that are underperforming. They will "balance" themselves in this way with time.
   2. And if it is for something that doesn't eventually get consumed into the full production chain (ie, military defense), then you can setup something that prioritizes it. It's not that hard. We do things like this all the time in other areas of the game that don't even need it.

And you know, as I think about it, all of this is technically true for trains, too. Think about the threads complaining about trains and the way they distribute/prioritize when you're consuming more than you can produce and the requests for things like round-robin to solve this instead of just simply expanding mining operations....



I don't want to see instantaneous fluid transmission in this game. That just feels wrong and doesn't go with anything else in this game (with the exception of power transmission). There may be a better way to do fluids that isn't that, but then we start hedging into performance tradeoff and time spent vs what is gained.

[coppercoil]

I don't want to see instantaneous fluid transmission in this game. That just feels wrong and doesn't go with anything else in this game (with the exception of power transmission).

I'm not sure if most of the players think the same. Would you say "you all play it wrong"?

[FuryoftheStars]

I don't want to see instantaneous fluid transmission in this game. That just feels wrong and doesn't go with anything else in this game (with the exception of power transmission).

I'm not sure if most of the players think the same. Would you say "you all play it wrong"?

Well, I wasn't speaking for everyone else with the section you quoted. I put extra line breaks in before it for a reason. That said, it does feel like an issue that is rarely brought up, and only ever by those that are seemingly into min-maxing, which I would dare say is not a majority of the player base. And I feel as if a strong argument could be made that instantaneous fluid transmission doesn't fit with the game is more than a simple opinion

Specifically to the point of "you all play it wrong": 1) you all are asking for mechanic changes to the game that have in-game solutions to, and 2) technically speaking, that distinction goes to the devs themselves:

This is the same answer as it was 6+ years ago. It's such an incredibly simple solution - I have no idea why people insist on making fluid splitters when the root issue is "You don't have enough fluids: make more"

[XT-248]

I will start with an example direct from one of the FFF posts.



From: https://factorio.com/blog/post/fff-402

Here, we observe a Gear Foundry efficiently filling two green belts with stacked gears. The blog shares the Foundry's details: "plus 2,500% craft speed and 4.33 craft per tick." Two fluid inputs are on the left side (visible in the preview window).


What interests me is that the fluid input is hidden behind the preview window, either intentionally or unintentionally.



So, in the interest of science, I will try to determine how much fluid a vanilla Factorio with that kind of overclocked speed would need in a second.

At approximately 1850% (32 beacons with two speed tier 3 modules and three speed tier 3 modules inside the Refinery), the necessary oil fluid intake goes from 20 units per second up to 370 units per second.

The standard pipe length between pumps is typically 17 sections, giving 1,200 units per second.

Ordinally, this would be sufficient to supply 60 refineries. At higher throughput, splitting the fluid movement between machines becomes necessary.

1200 fluid per second between pumps divided by 370 oil fluid intake for higher throughput refinery = ~3.24. Since we can not have a 1/4 of a machine, we have to round it up to four machines running at ~81% of the potential craft speed.

Alternatively, four-manifolds with overflow to an extra machine could keep all machines active at 100% throughput. The speed of the refineries could be dramatically reduced so that the pumps can keep up with the high throughput, or the throughput of pipes could be raised by shortening the length between pumps. All tweaks come with trade-offs, each with its own set of pros/cons.


Anyone who thinks the current pipe toolset is sufficient for a '+2,500% speed' machine needs to take a step back and re-examine their bottlenecks: fluid logistics.

I would hold off making assumptions until WUBE reveals more of the new fourth machines, which may be a 'new' petroleum-related machine.

[FuryoftheStars]

I will start with an example direct from one of the FFF posts.

[removed image for post size]

Here, we observe a Gear Foundry efficiently filling two green belts with stacked gears. The blog shares the Foundry's details: "plus 2,500% craft speed and 4.33 craft per tick." Two fluid inputs are on the left side (visible in the preview window).


What interests me is that the fluid input is hidden behind the preview window, either intentionally or unintentionally.



So, in the interest of science, I will try to determine how much fluid a vanilla Factorio with that kind of overclocked speed would need in a second.

At approximately 1850% (32 beacons with two speed tier 3 modules and three speed tier 3 modules inside the Refinery), the necessary oil fluid intake goes from 20 units per second up to 370 units per second.

The standard pipe length between pumps is typically 17 sections, giving 1,200 units per second.

Ordinally, this would be sufficient to supply 60 refineries. At higher throughput, splitting the fluid movement between machines becomes necessary.

1200 fluid per second between pumps divided by 370 oil fluid intake for higher throughput refinery = ~3.24. Since we can not have a 1/4 of a machine, we have to round it up to four machines running at ~81% of the potential craft speed.

Alternatively, four-manifolds with overflow to an extra machine could keep all machines active at 100% throughput. The speed of the refineries could be dramatically reduced so that the pumps can keep up with the high throughput, or the throughput of pipes could be raised by shortening the length between pumps. All tweaks come with trade-offs, each with its own set of pros/cons.


Anyone who thinks the current pipe toolset is sufficient for a '+2,500% speed' machine needs to take a step back and re-examine their bottlenecks: fluid logistics.

I would hold off making assumptions until WUBE reveals more of the new fourth machines, which may be a 'new' petroleum-related machine.

I really don't know what you're trying to get at here. This seems like more of an issue of how much fluid you can pump through a pipe than an issue of the fluid not dividing equally between the branches? I mean, you either have 4 machines running at ~81%, or you have 3 at 100% and one at ~24%. Either way, you're still consuming 1200 fluid/s and producing the same number of gears.

If your issue is with how quickly fluid pressure/rate of transfer through the pipes drops off, then yeah, sure, a (separate) discussion could be had there and that could be easy to balance and adjust without performance impact. But the OP of this thread, as I understood it, was a complaint in the way that fluid unevenly divides at splits, which is a different animal.

[mmmPI]

I will start with an example direct from one of the FFF posts.

IMO that's a terrible example to show that "current fluid mechanic is broken", since we have no idea if it is using or not the same mechanic than currently.

Another reason is that the original post is about the update order and unbalance splitting that occurs when lacking fluid, which is unrelated to the setup you use as example since there is no splitting occuring to feed a single foundry.

The standard pipe length between pumps is typically 17 sections, giving 1,200 units per second.

According to who ? why 17 ? seem to me quite the arbitrary number. Sure offshore pump can output only 1200 water per second, but you can group them an then use regular pumps to transfer 10 times faster to 12000 fluid/ second.

That would still be unrelated with the "problem" mentionned in OP though what you describe is related to max throughput of fluid which can be easily solved by adding parralelel input of fluid. It has nothing to do with fluid junction splitting unevenly in case of low fluid.

[Rseding91]

That machine is using 2 fluid per craft and as of writing that FF the fluid mechanics are the same as 1.1.

[FuryoftheStars]

The standard pipe length between pumps is typically 17 sections, giving 1,200 units per second.

According to who ? why 17 ? seem to me quite the arbitrary number.

That is the actual measured flow rate in game according to the wiki: https://wiki.factorio.com/Fluid_system#Pipelines

But everything thing else you said is true and I agree with.

[mmmPI]

The standard pipe length between pumps is typically 17 sections, giving 1,200 units per second.

According to who ? why 17 ? seem to me quite the arbitrary number.

That is the actual measured flow rate in game according to the wiki: https://wiki.factorio.com/Fluid_system#Pipelines

But everything thing else you said is true and I agree with.

But it doesn't make it a standard to me . I mean why not say 7 is the standard pipe length, because it gives 1500 fluid per second ? or 200 because you'd still get more than a 1000 fluid per second ?

Then the max throughput is not the same problem. If it is a problem. Anyway its not the same thing as mentionned in the OP.

And according to Rseding91 informations, 2 fluid per craft and the 4.33 craft per tick, it makes 8.66 fluid per tick requirement, which is 519 fluid per second i suppose that mean you could have 2 of those foundries at the end of a 50 pipes length, and still be fine for throughput.

[FuryoftheStars]

The standard pipe length between pumps is typically 17 sections, giving 1,200 units per second.

According to who ? why 17 ? seem to me quite the arbitrary number.

That is the actual measured flow rate in game according to the wiki: https://wiki.factorio.com/Fluid_system#Pipelines

But everything thing else you said is true and I agree with.

But it doesn't make it a standard to me . I mean why not say 7 is the standard pipe length, because it gives 1500 fluid per second ? or 200 because you'd still get more than a 1000 fluid per second ?

Then the max throughput is not the same problem. If it is a problem. Anyway its not the same thing as mentionned in the OP.

And according to Rseding91 informations, 2 fluid per craft and the 4.33 craft per tick, it makes 8.66 fluid per tick requirement, which is 519 fluid per second i suppose that mean you could have 2 of those foundries at the end of a 50 pipes length, and still be fine for throughput.

Yeah, sorry, it didn't cross my mind that you were specifically picking on the fact that XT was referring to 17 pipe segments as "standard" vs the fluid amount for that until after I had posted, but I decided to just leave it.

As for the rest, I fully agree and am not debating any of that. Personally, I found many of the numbers that XT picked as being rather arbitrary, but ultimately I didn't care because it seemed the point they were trying to make was unrelated to what this thread was even about.

[mmmPI]

I thought there could have been a misunderstanding between 1200 fluid per second for offshore pump and 12000 fluid per second for regular pump which could cause an unecessary worrying about max throughput of pipes and thought it could help to dispell the misunderstanding if one has been playing with only 17 long section of pipe thinking throughput couldn't be made better due to pump limitation when pumps have a limitation of 10 times the offshore pumps.

It's unrelated to the original problem but i thought players that have trouble with the fluid system may not be able to clearly identify the cause of them, and mistakenly add onto this topic, making explanations redundant sometimes help

[XT-248]

That machine is using 2 fluid per craft and as of writing that FF the fluid mechanics are the same as 1.1.

With the overclocked Foundry doing 4.33 crafts per tick, a tick is 1/60 of a second and consumes two fluid units per craft.

It is easy to obtain the necessary pipe throughput from those known numbers. 4.33 craft per tick * 60 ticks in a second * 2 fluid unit per craft = 519.6 fluid units per second.

Returning to the earlier example of the overclocked refinery, we can see that it provides a reasonably accurate approximation, even if it's off by a few hundred units.

-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

I will reiterate and expand on what Tertius said about building a factory from top to bottom.

Suppose I place a pipe network from remote oil outposts to train stations or refineries. In that case, I do not know beforehand if it would work up to a hypothetical capacity if it has several junctions (typical when you have oil patches offset from each other).

The same applies to petroleum gas, light oil, heavy oil, and lubricant oil to various degrees. Water deserves a mention here for long-distance pipe logistics to Nuclear Energy Heat Exchangers, and even then, water pipes barely count.


Part of the issue is that fluid can oscillate in any direction. Another side to the same problem is how the game decides which pipe gets priority. Now, higher throughput machines all drive the need for better pipe logistics.

Sure, producing more fluids works up to a point when the pipe becomes your bottleneck.

-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

As mentioned previously, I hit this limitation too quickly in modded Factorio 1.1.

It would take little effort to do so in vanilla Factorio 2.0 and Space Age with overclocked fluid input/output machines.

[mmmPI]

I have found a way to more than double the throughput from 1200 all the way up to 3000 fluid per second ! I think in no time it will become the new standard :

Throughput optimisation.jpg (344.29 KiB) Viewed 2024 times

0eNqV1dtqhDAQBuBXkbl2l5w8Xvct2lJ0DduAxqCxVMR3b3a96EqlP3tpmPkYzE9mobqdtBuM9VQu1OjxMhjnTW+ppMk5PURtZXUkGWNvtu6977vthAvGKCZz6e1I5etCo7naqr0pfnY6tBuvu1Bhq+725Uw4XEODbfQ3lXx9j0lbb7zRW//9Y/6wU1frIRTsO2Ny/Wi2wRa6AZlIz0lMM5UiSdk5Wdf4DyJ+kalzh0j+SMTUmEFftgpxAMrdVCffn65DP9nmkC728+3w9ABXT+DyP/xo8gT+CsWfAlN4QUrAC8qeuuX0GMkxkkGkwEgOEc6wUmAFZ18yrAiscKxIrAisKKxIrCRYUVjBwZUJVnByJU4ux9GVOLocZ1fi7AqcXYmzK3B2Fc6uwNlVOLtCPvVE3ZWwle57q3xYjDG1Va3DaqOXfqpbHfnP8C5fP93kI12Npp1DyZcexu2tzLnKCpHlacEkU+v6A3rPTAo=

Copy blueprint

The good thing is that it also prevent fluid from oscillating in a way difficult to predict. One such setup allow to feed up to 5 foundries working at 100%. It can even provide for assembly machine 3 with speed module and speed beacon all around, those consume around 2800 water per second when trying to make barrels but then it become difficult to extract all the barrels without clocking the inserters, the bottleneck is not the pipe x).

An even better thing i found out, the upper setup doesn't get bad when you extend it :

updated version.jpg (104.69 KiB) Viewed 2020 times

I have good hope that the previous standard of only straight pipes that was causing bottleneck has no more reason to be thanks to this discovery, i wonder if many players are aware of it.

Sure, producing more fluids works up to a point when the pipe becomes your bottleneck.

When the pipe is the bottleneck the trick is to add another pipe in parralel, here is an example :

double throughput again.jpg (1.91 MiB) Viewed 2024 times

This setup is using twice in parralel the proposed new standard to achieve twice the throughput compared to a single lane.

Which side do you think has priority ?

i'm kidding , there are more fluid production than consumption, so it doesn't matter, no machine are starved, no problem !