Optimized Steam Engine Setup

[Patric20878]

Um, I'm absolutely new to Factorio (currently trying to get blue science^^) but maybe I found something interesting for you.
About your problem with the boiler from the detector storing 5 coal, why not just use a normal inserter to get the coal in
and a fast inserter (just to make sure) to get the coal out again.
I just happened to read two topics at the same time. I'll post a link in my next posting, apparently I'm not allowed to do this in the first one.

The link wasn't really what I was looking for, but your post did make me realize I could just have 2 burner inserters feeding into the boiler, and 1 normal one taking them out. Electric inserters can't be used to feed the boiler because they can't start with no power. With the 2/1 modified setup like that, response time would be quickened to however long it takes for the inserter to pull coal out, along with the current coal that's burning. Such an obvious solution too, derp. Added you to credits, thanks!

Intresting topic. Though I don't agree that optimising for power/area is the best solution. If your steam setup is massive enough to need a compressed setup then you are going to do it on a designated location outside of your base where you have no real limitations on far out from the shore it reaches. If it takes twice as much area but is 5% shorter then it would be less unwieldy since you need less shore and it's easier to find a good locaion for it. A powerplant that is 1km will be hard to fit in anywhere even if it has almmost 0 area. So it's good that you found a way to compress it so it uses less space between the engines. But limiting it to 9 engines is pointless, imho. I would slap on one more steam engine at the end to get 10 in all rows. Using only 9 of them mean you need more rows of steam engines to reach the same amount of power, which is less compressed then optimal. I might try 25 steam engines in each row if the extra pumps needed don't use too much power.

Actually, if I'm building a steam setup at all, my base will be right next to it. I don't know about you, but I think everyone prefers building their starting base next to a large lake rather than half a mile from a small pond. And 25 steam engines each row doesn't even work, offshore pumps aren't that powerful. If my steam engine setup were massive, it'd obviously be on a massive lake. And what are you talking about, less compressed than optimal? You use MORE space to use 10 steam engines per row than I do to just build another row. You're assuming that the lake is limited but available land isn't, and frankly it could go either way, with trees, rocks, other lakes, and massive amounts of nearby spawner camps all giving it a good reason to be as compact as possible. Play on highest spawner settings and just watch how fast you run out of space. But in the case none of those apply, use Aru's variant, which uses 10 steam engines per row.

[kinnom]

25 per row is actually doable viewtopic.php?f=134&t=6066

[Patric20878]

That post is about pipe throughput, not offshore pump output. 25 steam engines per 3-tile row (yes, including offshore pumps) is not possible, because each offshore pump only supports 10! However, as offshore pumps are 2 tiles high instead of 3, it may be possible to arrange 15 steam engines per 3-tile row. This I will look into. The inserter setup will almost guaranteed use boiler chains (bad thing) if this is possible though, so depending on how bad it is, it may or may not be a good idea to do it.

[Qon]

Intresting topic. Though I don't agree that optimising for power/area is the best solution. If your steam setup is massive enough to need a compressed setup then you are going to do it on a designated location outside of your base where you have no real limitations on far out from the shore it reaches. If it takes twice as much area but is 5% shorter then it would be less unwieldy since you need less shore and it's easier to find a good locaion for it. A powerplant that is 1km will be hard to fit in anywhere even if it has almmost 0 area. So it's good that you found a way to compress it so it uses less space between the engines. But limiting it to 9 engines is pointless, imho. I would slap on one more steam engine at the end to get 10 in all rows. Using only 9 of them mean you need more rows of steam engines to reach the same amount of power, which is less compressed then optimal. I might try 25 steam engines in each row if the extra pumps needed don't use too much power.

Actually, if I'm building a steam setup at all, my base will be right next to it. I don't know about you, but I think everyone prefers building their starting base next to a large lake rather than half a mile from a small pond. And 25 steam engines each row doesn't even work, offshore pumps aren't that powerful. If my steam engine setup were massive, it'd obviously be on a massive lake. And what are you talking about, less compressed than optimal? You use MORE space to use 10 steam engines per row than I do to just build another row. You're assuming that the lake is limited but available land isn't, and frankly it could go either way, with trees, rocks, other lakes, and massive amounts of nearby spawner camps all giving it a good reason to be as compact as possible. Play on highest spawner settings and just watch how fast you run out of space. But in the case none of those apply, use Aru's variant, which uses 10 steam engines per row.

I'm talking about end game steam setups for GW of power. The situation is a bit different then.
And like I said, it doesn't matter if it uses more area, the important thing is compressing it shore length wise because you are not building it inside your base anyways if it's big enough. Rocks and trees are removable, even on a massive scale once you get to fully upgraded contruction robots. Spawners camps on max settings are weaker than the player at the end game, so not a problem either. The shore can support 15 steam engines per row. I made a nice design for 15 steam engines every 3rd tile average. Thats 58% more power/shore length than 9.5 steam engines.

More steam

SteamPowerOff.png (4.35 MiB) Viewed 5640 times

Sadly the shore can't support more than 15 steam engines.

That post is about pipe throughput, not offshore pump output. 25 steam engines per 3-tile row (yes, including offshore pumps) is not possible, because each offshore pump only supports 10! However, as offshore pumps are 2 tiles high instead of 3, it may be possible to arrange 15 steam engines per 3-tile row. This I will look into. The inserter setup will almost guaranteed use boiler chains (bad thing) if this is possible though, so depending on how bad it is, it may or may not be a good idea to do it.

Yes, 25 is not possible with off shore pumps because of their size.

And no boiler chains are needed for that many steam engines

[Patric20878]

That's a seriously incompact solution you have there. Area DOES matter, even if you try to convince yourself it doesn't. If you're using "outside your base" as an argument, just build everything outside your base, including your base, then nothing matters for size, including your base. The whole point is making the best possible setup in all aspects without cutting corners, like making a suboptimal solution by arbitrarily dismissing factors as unimportant. Not only that, but half your rows right side is empty to make space for a 2nd coal belt branch. Very valid aspect maximizing power per shore length as well, but it does NOT replace power/area optimization. I'll need to see if 15 per row is possible without dropping power/area massively. No one said the two are mutually exclusive, so it may well be possible that increasing steam engines per row increases power/area as well.

[Qon]

That's a seriously incompact solution you have there. Area DOES matter, even if you try to convince yourself it doesn't. If you're using "outside your base" as an argument, just build everything outside your base, including your base, then nothing matters for size, including your base. The whole point is making the best possible setup in all aspects without cutting corners, like making a suboptimal solution by arbitrarily dismissing factors as unimportant. Not only that, but half your rows right side is empty to make space for a 2nd coal belt branch. Very valid aspect maximizing power per shore length as well, but it does NOT replace power/area optimization. I'll need to see if 15 per row is possible without dropping power/area massively.

Seriously, area DOES matter. I'm not saying it is completely irrelevant. All I'm saying is that I would prefer my way of doing it. And yes, if your base grows big enough then you have to split it into multiple bases with trains running between them. I would say that space is the most precious resource in the game. But for a steam engine setup that is outside your base I want to limit the shore length required, because it matters too. I'm not arbitrarily dismissing factors, and I'm not saying it is better in all cases. I haven't completely disregarded area either. My setup is fairly area efficient, though not as area efficient as yours. My setup uses 2x more area exactly (if you move the left underground pipes one tile to the left and the right coal belt one tile to the right) and gives 58%~ more power.

Which is a fair tradeoff because if you have biter problems then what you need is not area optimisation, it's perimeter optimisation. You don't defend area, you defend the perimeter. The double steam engine length is negligible in terms of extra turrets needed for a big power plant. My setup has a shorter perimiter because it doesn't stretch out the edge length needlessly. My setup uses only only 63% of the shore length of your setup. That means you need only 63% as many turrets and walls and so on to defend you power plant. And it also mean you need to clear out 37% less biters, because you don't clear out the exact are you place your engines at. You also have to clear out a fair bit around your setup so you aren't attacked constantly, right? The native base expansion distance is 160 tile radious, which I assume you would at minimum also clear out. You then go from 231 (67 steam engines, 4 for laser wall + stone wall (minimum defence, no roboport repairs or anything) + 160 expansion radious) to 297 tiles from shore to clear (+67 because my setup uses that much more tiles from shore). That is 29% more land to clear from biters for 58% more power for that land. And you could fit roboports in my design. 1.29 * 0.63 = 0.81. That means if you use my design you need to clear 19% less are for the same amount of power! That is assuming you clear out nests close enough to not be able to expand back (I would do that and probably more), that you aren't intrested building any factories at the end of the steam engines (might not apply to all) and that your steam power plant is so big that the extra turrets at the ends is negligible compared to the long line of turrets at the end of steam engines.

[Patric20878]

You don't get it do you? Power/area optimization is DIRECTLY linked to perimeter optimization. If anything, making a design more rectangular instead of more square INCREASES perimeter, which is exactly what your design is doing. This is elementary school math that I really don't feel like teaching. And enough said, as I've already came up with a solution that incorporates both 14/15 rows AND has slightly higher power/area. All that needs to be done is seeing if there's a way to optimize that. Also, if you're still around, Aru, I could use your math that determines how much this affects stackability, if it does at all. I don't think it positively affects it but I'm not completely sure either.

[Qon]

You don't get it do you? Power/area optimization is DIRECTLY linked to perimeter optimization.

No, you can increase area and decrease perimeter simultanously.

If anything, making a design more rectangular instead of more square INCREASES perimeter

Yes. Which is exactly why I made it more rectanglular square when stacked.

which is exactly what your design is doing.

No. The design as imaged stacks vertically with the shore to the left...

This is elementary school math that I really don't feel like teaching.

I feel the same. Good that we agree on something...

[Patric20878]

If anything, making a design more rectangular instead of more square INCREASES perimeter

Yes. Which is exactly why I made it more rectanglular when stacked.

Which is a fair tradeoff because if you have biter problems then what you need is not area optimisation, it's perimeter optimisation. You don't defend area, you defend the perimeter. The double steam engine length is negligible in terms of extra turrets needed for a big power plant. My setup has a shorter perimiter because it doesn't stretch out the edge length needlessly.

Am I missing something or does this completely contradict your post about reducing perimeter needed to defend the engines?

which is exactly what your design is doing.

No. The design as imaged stacks vertically with the shore to the left...

Increasing perimeter by making the design more rectangular and have less power/area is what you're doing.

In optimization phase now, the offshore pumps do fit, and the line of pipes it requires still allows the design to have higher power/area than the 10/9 setup. AND it's 15/15 too. Looks like the geometry of this will be able to incorporate both higher power/area and maximized engines/row, as 14/15 doesn't compact nearly as neatly as 9/10 does.

[Qon]

If anything, making a design more rectangular instead of more square INCREASES perimeter

Yes. Which is exactly why I made it more rectanglular when stacked.

Which is a fair tradeoff because if you have biter problems then what you need is not area optimisation, it's perimeter optimisation. You don't defend area, you defend the perimeter. The double steam engine length is negligible in terms of extra turrets needed for a big power plant. My setup has a shorter perimiter because it doesn't stretch out the edge length needlessly.

Am I missing something or does this completely contradict your post about reducing perimeter needed to defend the engines?

which is exactly what your design is doing.

No. The design as imaged stacks vertically with the shore to the left...

Increasing perimeter by making the design more rectangular and have less power/area is what you're doing.

In optimization phase now, the offshore pumps do fit, and the line of pipes it requires still allows the design to have higher power/area than the 10/9 setup. AND it's 15/15 too. Looks like the geometry of this will be able to incorporate both higher power/area and maximized engines/row, as 14/15 doesn't compact nearly as neatly as 9/10 does.

I meant more square. Ooops wrong word.

[Patric20878]

Anyways, any discussion of what is more and less important is all pointless now, because the new geometry allows for 15 steam engines each row without actually increasing width. Pretty great. Now uses 15 steam engines per row instead of 10/9, and not only does it maximize engines per row, but even increases power/area by around 3% or so. Maximizing to 15 engines per row will remove the minor disadvantage mine had not using the full supported engines per row, thanks to incredibly lucky geometry that results in every other row being shorter than the other, which is due entirely to 15 being an odd, instead of even number. Original post will be updated once this is finalized, crediting you for the find.

[Qon]

Anyways, any discussion of what is more and less important is all pointless now, because the new geometry allows for 15 steam engines each row without actually increasing width. Pretty great. Now uses 15 steam engines per row instead of 10/9, which means it not only uses the maximizes engines per row, but even increases power/area by around 3% or so. Original post will be updated once this is finalized, crediting you for the find.

I did try to compress it further by doing double rows of engines and lots of other creative placement. But then I didn't reach maximum theoretical output that the straight simple lines allow. So just make sure your design actually reaches 510 kW*number of steam engines. q:

Thanks for credits, and thank you for making such an interesting thread and such a good base design to build on for me

Eager to see what you came up with, because I know my proof of concept can be improved further.

[Patric20878]

Ran into complications: the engines must start up at least once disconnected before it can operate normally, otherwise it outputs 43.4 MW instead of 45.9 MW, on 90 steam engines and 21 boilers per row. I don't get it - by math, 20 boilers (7800 KW) should be enough for 15 steam engines (7650), but that only outputs 45.6 MW everytime. More boilers don't even help. The reason has to be water pressure - that would explain why more boilers don't help and why it must start up disconnected once - both are related to water pressure. This means there's an upper limit on just how long a pipe can be before small pumps are required, and small pumps definitely can't go into the standard variant, since it requires tech.

And also, the water battery variant works terrible with this. But if the reason is indeed water pressure, relocating the small pumps to after the boilers should make a big difference. But otherwise, 15 engines per row is completely compatible with no boiler chains!

[Qon]

Take a look at the bottom of OP in the long distance pump thread.
He there shows 25 engines on a pump. I think it uses at least 2MW for the pumps though! But if all you want is 15 then you will need less small pumps (3 in parallell) and not as often as once every 4 boilers.
But those are the reasons I avoided chaining all the engines, no energy wasted on pumps and less hassle with water pressure. But maybe it's possible to make it that way. Maybe it's enough with just one small pump placed correctly.
I'll take a look at making my setup compressed again later. Maybe I can do it without losing efficiency. It's tricky.

[BlakeMW]

Ran into complications: the engines must start up at least once disconnected before it can operate normally, otherwise it outputs 43.4 MW instead of 45.9 MW, on 90 steam engines and 21 boilers per row.

If I understand correctly this is a general problem. If you take a bog standard 1/14/10 empty of water and hook it up under full load, then turn on the offshore pump, it will sometimes fail to reach it's potential, ever. Unless a steam engine system is given some slack time to fill with water OR has more than 1 offshore pump to 10 steam engines this problem can occur. This is a potential problem with backup steam which uses a direct circuit network connection to the offshore pumps, if the system is triggered under full load it'll potentially run only about 95%.

In general whether or not this will happen is heavily sensitive to placement order and also direction. Altough placement order is a little bit of a dark art, if you place everything in the same order as the water flow (first the offshore pumps then the boilers then the steam engines), then the water fill flow through a nearly empty system is much faster (about 2.5x faster) than if the placement order is "retrograde" (against the flow), I believe the high speed flow through a nearly empty system can sometimes cause problems with water passing through the boilers too fast.

On the primary topic of this thread, I'm terribly happy with my design I title "Smogatron 160":

Smogatron 160

Blueprint

In terms of compactness there's only a little slack - I choose to use electric inserters so energy isn't wasted on burners. If you tile these there is 1 vertical tile of "slack". The madness or genius is using a block of 16x10 steam engines, but having the strings 16 steam engines long instead of 10 long, this matches the boiler strings really well being a little better than boiler strings 15 long. The design can also be easily cut in half, it's really two 80 steam engine power plants sharing one belt, the two halves do not share pipes.

In my testing 80 steam engines on one shared pipe is pretty much the limit - try to do more and the fluid simulation starts to lose it. What tends to happen if you try to add even more boiler and steam engine strings to the same pipe system is the directional flow bias tends to lead to some parts of the pipe saturating and the water from some of the boiler strings can no longer reach some of the engines. This can be remedied to some degree through the use of short-circuit underground pipes. For example the following design is a 10/140/100 using a shared pipe system which is able to operate under full load at 100% efficiency:

Shared Pipe 10-140-100

Note carefully the pipe-to-ground at the end of the bottommost boiler string. Without this short-circuit pipe-to-ground about 5 of the boilers are inactive. The short-circuit helps the water move north.

This design also works fine with a block of 5x20 steam engines, making the steam engine block more slender than the boiler block. Good for coast line optimization, but not for perimeter optimization.

There is a basic limit in how big power plant blocks can be because of the rate at which belts and inserters can deliver fuel. 140 boilers is close to the limit for a single lane of coal of an express belt. Once you go over that limit you can no longer flip the design and share a belt. It is hard to say what 0.13 will bring in terms of the inserter stack bonus nerf, and the rapid inserter. This will absolutely nerf burner inserters for boiler blocks and change the absolute limit of how large boiler blocks can get depending on exactly how the rapid inserter behaves when feeding between boilers.

[Patric20878]

In general whether or not this will happen is heavily sensitive to placement order and also direction. Altough placement order is a little bit of a dark art, if you place everything in the same order as the water flow (first the offshore pumps then the boilers then the steam engines), then the water fill flow through a nearly empty system is much faster (about 2.5x faster) than if the placement order is "retrograde" (against the flow), I believe the high speed flow through a nearly empty system can sometimes cause problems with water passing through the boilers too fast.

This along with the pipe to ground thing to short circuit something, break it down please. What empty system? Shared pipe?

And why doesn't my current design shown in original post not have that problem of not operating at full potential? Even though the water battery variant starts off less at less than full load, it recovers as it fills up the tanks with water. The issue with the one I'm developing is that in the 20 boiler one, some steam engines don't have 100 C water for reasons I don't comphrehend, and in the 21 boiler one, water level is too low in some steam engines. And I observed that when linking 3 offshore pumps into one pipe, the north one always seems to have the most water. And it's completely symmetrical too. What the heck is happening?

[BlakeMW]

Well the factorio fluid simulation is a bit hard to pin down, it's like a cellular automa although with some performance optimizations.

There is a definite flow bias in terms of placement order. Fluid elements are simulated in the order they are placed in - so each tick every fluid element will be processed. A really clear example of this happens in "hot water mods" such as the Oil Burner, in these mods there is an assembler which outputs water. The water is outputed into a fluid box. Then the mod script heats the water in the fluidBox, because in 0.12 there is no way to generate hot water. This is where placement order comes in, if the Oil Burner is placed before the outlet pipe, the water will flow out of the fluidBox before the script gets a chance to heat it leading to lukewarm output. On the other hand if the OilBurner is placed after the outlet pipe is placed, the script will heat the water every time guaranteed. Also direction matters too! If the OilBurner faces north or east, it will heat the water to 100 degrees regardless of palcement order. If it faces south or west it will only heat the water to 100 degrees if it placed after the outlet pipe. More generally if the direction and placement order are correctly aligned water can almost teleport through pipe systems, as the fluid simulation moves it from pipe to pipe.

So both the direction and placement order effects the simulation. A setup which works perfectly when placed north to south may not work perfectly when placed south to north. And of course, placement order can also effect whether it works perfectly or not. As a general rule these wont cause more than 5-10% degradation of performance, and often it'll only be 1-2% which is quite acceptable to normal people (maybe not to perfectionists)... Also these effects only come into play when you are close to the limits of the system, a system which is nearing the limit of flow a single pipe can handle can manifest this degradation, but a system where pipes are only 1/3rd full generally wont. A system which is both close to the limits and started empty is most likely to manifest degraded performance. In general it's best to not be too close to the limits, this is one reason why 1:14 pump:boiler works really well as a building block for large powerplants, 1:14 is not even close to the limits of a pipe, it gives lots of slack to work with further down the line, with the steam engines and connecting pipes.

Now, as for short circuit pipes. Pipes can bottleneck if too much liquid tries to go through them. Happily the game is able to use parallel pipes fairly well to allow a greater flow. When water has the choice of two directions the split is not perfect, like if you were to fork and rejoin a pipe, perhaps 60% of the water will go down one path, and 40% down the other path, this makes fork then join a very bad strategy for the boiler, if you have two strings of boilers and need a 50% split, but the water splits 60/40, the result is too much water goes down one string to be heated and the water is only lukewarm, and too little water goes down the other and it's fully heated but some of the boilers go unused. But going parallel works well for flow balancing, in my 10/140/100 design there's no fork-then-join problem, it uses the safe join-then-fork pattern. But 1 pipe is inadequate for flow balancing purposes, too much of the water tries to go in one direction and the pipe bottlenecks, this shouldn't happen but it does because of directional and placement order flow biases. But as we know if there's enough excess capacity those biases stop mattering, that's where the parallel pipe comes in. In that case I use an underground pipe as a parallel pipe, it allows liquid to directly teleport from the bottom boiler string to the upper steam engines, this reduces the pressure on the primary pipe, freeing up some slack and allowing it to successfully balance the flow. Underground pipes are best because they are liquid teleporters (or wormholes), but you can also use standard pipes in parallel and it works pretty well. You can't double capacity with parallel pipes, but you can probably get a 40% increase in capacity which is often enough.

[Patric20878]

Well that's bad. If placement order affects this, why doesn't replacing offshore pumps and pipes make them split more evenly?

I may have to just post the imperfect version first. Accounting for all this is quite new to me.

[Aru]

You don't get it do you? Power/area optimization is DIRECTLY linked to perimeter optimization. If anything, making a design more rectangular instead of more square INCREASES perimeter, which is exactly what your design is doing. This is elementary school math that I really don't feel like teaching. And enough said, as I've already came up with a solution that incorporates both 14/15 rows AND has slightly higher power/area. All that needs to be done is seeing if there's a way to optimize that. Also, if you're still around, Aru, I could use your math that determines how much this affects stackability, if it does at all. I don't think it positively affects it but I'm not completely sure either.

I'm not certain what you're asking for... but, if you take a narrower rectangle with lower density, and you tiled both until they're roughly square-shaped and the same size, so that they have similar perimeter, the denser one is just as much more per perimeter as it is per area.

[Patric20878]

Oh, I meant how much stacking it in rows of 15 engines each affects stackability. Wasn't talking about perimeter. Would the main belt still only be able to support 10ish rows or would it support more?

Also, I want to confirm if I'm using your quadratic equation correctly. If I have 15 steam engines and 20 boilers, each row would require 2.16 burner inserters to prevent slowdown right?

And also, I think all burner inserters would have to be placed on the closest row to the coal source to prevent slowdown to the fullest. However, the fact that coal comes much slower shadows the slowdown from not having enough burner inserters, because it takes quite a while for all boilers to fill up with coal on a normal belt. Can you factor this into your math and recalculate? It would appear that it's possible for just one burner inserter per row to be enough to "prevent" slowdown, because any amount of slowdown is faster than the time it takes to fill boilers.