Jump to content

Extreme Liquid Duplicator - 36kg/s Super Coolant Geyser

Recommended Posts

36kg/s super coolant geyser :

Drips with another liquid sometimes causes liquid duplication. 


1.How to Duplicate Liquid

Mesh tile version is simple and stable.

A liquid vent with a mesh tile converts a drip into a liquid tile immediately. If the vent is under light liquid, it seems to make not only a converted liquid tile but also a flowing liquid tile. Increasing rate of suuper coolant is 1250g/s.


2.High Throughput Duplication

Dropping heavy liquid on light liquid drips (20g/s) also cause duplication.

Throughput of this version is not limited by vent. The escher waterfall duplicates liquid very fast.

Extreme version's save file is attached. I don't know detail of hydrodynamics. There is probably room for improvement.


Link to comment
Share on other sites

@nneko25, figured you would enjoy seeing this. 


It generated 12.4 tons of supercoolant in one cycle, or about 20.66kg/sec. My guess was that I could get around 23.4kg/s with this duplicator, based off the viscosity of supercoolant. The beads fall in chunks of 75kg every 0.4s (some are smaller, which is why 23.4kg/s was a max). When they hit the duplication bug, exactly 1/8th of each bead is duplicated, resulting in 75/8 = 9.375 kg per 0.4s (or 23.4375 kg/s). 

The bug does not require a vent or a mesh tile.  What is required to trigger the bug is a bead of liquid that (1) will fall down fully and (2) can push another liquid left while leaving a vacuum between the two.  

I used a bead configuration to force the beads to fall every 0.4 seconds (that's the crude oil stuff up top).  The two tiles above the crude oil are needed to prevent a waterfall from forming (which stops the duplication).  From there, I used an escher waterfall to pump the liquid back up as quickly as it is created.  

 At this point, I think it's time to do the following:

  1. Make it more compact (min).
  2. Increase duplication to theoretical maximum (max).
  3. Duplicate gold (gotta aim for alchemy always. :)
  4. Attach a mini version into a SPOM (with built in cooling occurring at the duplication point).
  5. Power an army of turbines that fill the map (lava duplication?) 
  6. Hopefully draw enough attention that it gets patched out quite rapidly.   

The possibilities are essentially endless.  I'll put more details about the exact nature of the bug in the bug report. Thank you for posting a report there.  Have fun all with this beast.  @Saturnus, if you want a fun min/max problem, I think we've got enough here for you to construct another glorious cube. @ghkbrew, this one might up your alley as well. The fun with this bug will probably be short lived. 

Well, my initial post above has the wrong details.  Here is what's happening (from the bug report).

The bug happens at the bottom of the beading supercoolant.  There are three tiles on that same level.  The left most tile has 60g of water, enough to force the water to flow right when it can.  It appears that there is a race condition for control of the middle tile.  


Here is the sequence of events, every 4 ticks (0.8 seconds)

  1. Vacuum in middle tile
  2. 1/4 of water from left cell pushes into middle cell. 
  3. Water stays there for one tick, while the supercoolant drops down. 
  4. Duplication of exactly 1/8 of the liquid supercoolant to the right appears in the center tile. THIS IS THE BUG. 

So in this contraption, the beads fall around 132 kg.  The duplication occurs every 0.8s (4 ticks), netting 132/0.8 = 20.625 kg/s of free liquid.  I'm sure I can find ways to increase this rate. :) I hope this helps isolate the bug.  

Link to comment
Share on other sites

I don't know if I can make it any simpler than this. The tile above the vent doesn't have to be vacuum (it can't be the same liquid as leaving the vent). 


In this setup, I am using 60g of water on both sides of the vent. This is enough liquid that it won't stay in one tile if it can flow, but not enough liquid to spill over into three tiles. So the vent tile will stay vacuum except for the 1 tick when liquid leaves the vent.  The next tick afterwards, the water is pushed left and right, while 1250g magically appears both left and right of the vent.

Clearly the contraption above doesn't yet have a way to extract the excess liquid.  I figure adding a hydro senor, along with a liquid shutoff somewhere, should be sufficient. 

For those who demand 4 tile high rooms, I've failed you.  Sorry.  I'm sure you can find a way to adapt this. :) 

Well, with 60g water on each end, and 60g water between the two vents this produces a steady 5kg/s supercoolant.


If you are concerned about the 240W needed to run the pumps, then you can build a more elaborate version that uses escher water falls to bead the liquid onto the vents. :)  I think I'll stop worrying about infinite liquid storage. Who needs it when we have controllable infinite liquid generation. To generate water, I'll just use this to duplicate salt water and then run that through a desalinator. I've got to play with heat now, and see if the duplicated temperature comes from the vent tile, or the liquid below. If it comes from the vent, then we get a nice boost to any temperature modifications. 

Link to comment
Share on other sites

10 hours ago, mathmanican said:

To generate water, I'll just use this to duplicate salt water and then run that through a desalinator.

If liquids generated are at a constant temperature, then I vote for brine.

Link to comment
Share on other sites

8 minutes ago, JRup said:

If liquids generated are at a constant temperature, then I vote for brine.

You can generate any liquid, at any temperature. Magma, liquid niobium, liquid hydrogen, you name it.

Link to comment
Share on other sites

Visco-gel (or naphtha) version duplicates 2x or more faster. Visco-gel return in 1 tick. Duplicated liquid are added to input and duplicated again. Both direction version works.


Mass of visco-gel must be smaller than input and higher than duplicated (1/8 of input). Half of input seems good. (ex: 5kg for 10kg/s vent)

Max duplication rate is 1/8 viscosity / tick. Over viscosity input cause unduplicated side flow and duplication efficiency drops. Max duplication rate is 1/8 maxmass / tick. Upper liquid tile and lower void tile cause over viscosity duplication. Max duplication rate is unlimited. High pressure liquid duplication also works.

Vent version (For water duplication) :


Visco-gel duplication also works with naphtha. (250g/s, very slow)


detail :



Link to comment
Share on other sites

3 hours ago, nneko25 said:

duplicates 2x or more faster

Nice find.  I tried the same setup with insulated tiles, and got no duplication. I though the tile under the duplicated spot had to be the same liquid, but airflow might just do it. I think the airflow tiles here make a difference.  I'll explore that next chance I get. 

Link to comment
Share on other sites

2 hours ago, mathmanican said:

I think the airflow tiles here make a difference.

@nneko25, the airflow has nothing to do with it.  The key was having the vent/mesh tile one tile higher. This duplicates crude oil using petro. 


And here we duplicate petro using crude oil. 


The duplication rate using the two configurations above is halved, because of timing on ticks being off with liquid location.  I have no valves running, rather just dumping 10kg/s (but the tile under the mesh must have liquid in it). 

As for the amount of liquid on the side, we just need to make sure that the liquid won't flow to the middle.  For crude and petro, you can use 1kg with no issues.  

This configuration will duplicate crude at the full rate (2.5kg/s), as the crude doesn't have to be pushed back inwards just falls straight down. The mesh tile can be dropped one with the same result. The tiles under the petro can be removed also.  


Link to comment
Share on other sites

@mathmanican This is the minimum version. Vent input (slow, 10kg/s) requires 5kg visco-gel (or naphtha) to prevent re-duplication of only duplicated liquid. (Since there is no water pressure sensor, it will stop when the water level rises.)


1 tick return don't cause 2x faster with slow vent. This is useful for too light liquid (water) or vertical 2 vents. Re-duplication cause a little faster with 2 vents.


Link to comment
Share on other sites

Perhaps this is horizontal pressure equalization of multiple liquid. This process cause 1/8 of mass flow. Ultra high pressure liquid can move horizontal with another liquid pressing. Visco-gel also flows.


Link to comment
Share on other sites

So the 1/8th rule is a new one that I haven't explored before.  Apparently when liquids pull away from another liquid, they drag 1/8th of the touching liquid with them.  I'm guessing this is to model how fluids like to stick together (goopy).  

So we can suck off 1/8th a tile of naptha if we can force the liquid on one side to vacate.  This would be where the 1/8th duplication is coming from as well (normal single element liquid flow uses 1/4th the mass).  Naptha and viscogel work nicely for duplication because you don't end up using regular flow at all to do the duplication. As such, there is never a tick where vacuum appears, hence you can get full uptime on the duplication process and the bead pump duplicates every other tick (rather than every 4th tick).

Very interesting fluid mechanics properties. I stopped exploring after wearing myself out on data collection for single liquids. I think I'll let the dust settle on a fix to this bug before diving fully into more complete testing.  This is quite a crazy bug. 


This is now just silly. 


The petro stuff is just leftovers from an old build.  The valves are needed to jump start the system, but once it's running they are set to 0kg.  

Link to comment
Share on other sites

On 5/5/2021 at 7:15 AM, mathmanican said:

The bug happens at the bottom of the beading supercoolant.  There are three tiles on that same level.  The left most tile has 60g of water, enough to force the water to flow right when it can.  It appears that there is a race condition for control of the middle tile.  

That sounds suspiciously like how Alchemy used to work, only this time with liquids. 

Link to comment
Share on other sites


This topic is now archived and is closed to further replies.

Please be aware that the content of this thread may be outdated and no longer applicable.

  • Create New...