"Heat Doubling" Bug Hunt

[ghkbrew]

First off, I've been trying to prove to myself @mathmanican 's explanation and example of heat doubling by freeze/thaw cycling shouldn't work.  The reasoning flaw in his description is that between steps 5 and 6 600K solid lead is transformed to 600K molten lead.  This transformation actually requires heating the solid lead to 603K, and the molten lead will appear at 601.5K: a net lose of 1.5K worth of heat.  Exactly balancing the gains from steps 1-4.

However, he also provides a working machine... which makes me question my knowledge of ONI physics.

To resolve my doubts, I built this contraption, which freezes and thaws lead at 500kg/s at very, very close to ideal temperatures.

 

To my great delight, it was initially very slightly power negative (as would be expected by running a steam turbine off of a super coolant aqua tuner)

However I soon noticed that this machine was slowly accumulating liquid lead in the reservoir...  I intended to duplicate heat, but ended up duplicating lead.  Close I guess.

To complicate things more, after adding the lead overflow storage in the upper left, the machine became power positive. I think that can be accounted for simply because I'm melting less lead while freezing the same amount.

In conclusion, please help me figure out what the heck is happening here.

 

Lead Cycler.sav

[mathmanican]

3 hours ago, ghkbrew said:

This transformation actually requires heating the solid lead to 603K

This is the flaw.  It does not require heating to 603K when the solid lead is in debri form.  When solids are in debri form, they instantly change to liquid when they hit the phase change temp.  There is no extra 3K needed to phase change. If the lead were a solid tile, then it would have to heat up. 

Here is a test I just ran to confirm this. 

Nothing in this room ever passes 602K.  I had to get the hydrogen to 602K so that I could get the door to 601K (as I have to reach 600.7K to phase change).  I started with lead debris inside the door, and the drop out at exactly 600.7K (exactly).  

The only thing that you have to do to double energy, using the contraption you pointed to, is keep the cooling as close to 599.2K as possible.  Every degree you overshoot is something you have to fully reheat. 

3 hours ago, ghkbrew said:

slowly accumulating liquid lead in the reservoir

I wonder if they reintroduced the matter duplication bug.  If so, then we can make manmade geysers again.  I'm gonna have a look. 

Spoiler

 

 

@ghkbrew, I also opened up your file (gotta diagnose the extra matter issue).  You have lead appearing at exactly 600.7K below your mesh tiles, as expected.  The cool thing about the mesh tile approach is that no heat interaction occurs with the doors above, provide you never have too much lead fall out. 

3 hours ago, ghkbrew said:

accumulating liquid lead in the reservoir

What I saw was that your device was not reheating the lead fast enough to melt fully, so you would get large build ups.  Once the lead touches the door, you loose benefits.  You end up heating up stuff you are going too cool.  A key piece is that you make sure the liquid NEVER stays in the mesh tiles (touching the doors).  

Let me know if you can replicate this. Maybe it is platform dependent (doubt it).  I'm on Ubuntu Linux 18.04.4

[mathmanican]

3 hours ago, ghkbrew said:

overflow storage in the upper left

On other issue that your build could easily fall prey to is the debri bug (though not sure). Here is a copy of the machine I made yesterday which uses aluminum, rather than lead.  The downside to aluminum is that it freezes to solid at any temp above 160kg (just checked 159, 160, and 161 several times), whereas does not form solid tiles at 500kg/s (lead form solid tiles when you pass 1600kg - you get debri at 1600kg but solid tiles at 1601kg).  Since lead has about 1/6 the SHC as aluminum, but aluminum has about 1/3 max flow rate, I can only get twice as much free power out of abusing SHC from aluminum than I can get from lead.

Here is my most recent abomination on this topic, the aluminum phase change heat doubling experiment. Have fun. 

AluminumPCHeatDoubling.sav

[0xFADE]

In my regolith melter I ran the regolith along molten hot glass and it would melt off at that temperature, then I would accumulate enough of it to drop in to a cooling chamber to solidify through 2 mesh tiles.  If I directly dropped it in instead of buffering it the output temperature of the igneous would start dropping for some reason(even though the regolith would always melt off at the same temp).

Anyway, that might improve the melter portion.

of that design?

[mathmanican]

If anything, this isn't really a bug, rather a way to skip the phase change hysteresis.  You can get the same thing by cooling gasses down inside pipes , or heating/cooling liquids inside pipes, if you use the 1/10 capacity issue. So basically it's the same as moving 20kg/s on a conveyor rail, except that with solids, we can stack em in huge piles, melt them in doors, and completely skip the phase change issues. 

So it actually seems somewhat consistent, to me, with the rest of the world.  Stuff in pipes seems to behave just like debris, and hence we get to skip completely phase change overages, if that's how we want to design something. 

[ghkbrew]

31 minutes ago, mathmanican said:

When solids are in debri form, the instantly change to liquid when they hit the phase change temp.

What! Why did noone tell me this?  Ok, so my master plan is a go after all.  Here's my thought, you can double heat at the melting point of basically every solid in the game.  If you chain them together, it's multiplicative... chaining salt, aluminum and lead would theoretically give 8x the heat produced by a thermium aquatuner. If you use a refinery you could add iron, igneous rock, copper and gold = 2^6 times heat production.

33 minutes ago, mathmanican said:

I wonder if they reintroduced the matter duplication bug. 

Update: the lead duplication bug is runspeed dependent.  It's activating the overflow about every other cycle on super-speed, but not once in over 100 cycles at 3x.  So a bug, but not one you it during regular gameplay.

[Lifegrow]

1 minute ago, ghkbrew said:

Update: the lead duplication bug is runspeed dependent.  It's activating the overflow about every other cycle on super-speed, but not once in over 100 cycles at 3x.  So a bug, but not one you it during regular gameplay.

Never test your builds on super speed, especially with autosave enabled.

 

[mathmanican]

30 minutes ago, ghkbrew said:

  If you chain them together, it's multiplicative

Yep. I'm excited to see someone else get excited about this. Each material has its own limitations. To truly use all the energy from this you will have a lot of fun balancing to perform.

[ghkbrew]

53 minutes ago, 0xFADE said:

that might improve the melter portion of that design?

Here's my regolith melter design.

The igneous rock forms at the tip of the lava blade.  I avoid the debris temperature reset bug, by having the two auto-sweepers grab fresh igneous rock immediately after forming and put into a infinite storage pile for buffering.

Side note: I'm pretty proud of this design, because it is basically perfectly efficient.  All of the heat added to the lava is used to melt regolith.  No cooler at all.  And it will gracefully ramp up to 20kg/s from a cold start.  I'm adding tiny amounts of hot lava to the reservoir for reheating, but it could easily be adapted to use a glass forge.

1 hour ago, mathmanican said:

On other issue that your build could easily fall prey to is the debri bug

You mean the debris temperature reset bug I had fun abusing this weekend? I'm pretty sure the mesh tile cooler / solid bypass completely avoids that issue.

 

1 hour ago, mathmanican said:

this isn't really a bug, rather a way to skip the phase change hysteresis.

It's not about avoiding the hysteresis and changing exactly at the freeze/melt point. It's the fact that the 1.5C temperature rebound isn't applied.  That's what makes it an unbalanced cycle.  There are plenty of unbalanced cycles in the game that give free resources, (lot's with critters, oil well -> petroleum boiler -> petro gen, ethanol and salt gas SHC loops). Most are pretty obviously intentional.  I think it's a little unclear if this one is an intended mechanic or not, but failing a bug fix I'm gonna assume it's fair game.

 

1 hour ago, mathmanican said:

You can get the same thing by cooling gasses down inside pipes , or heating/cooling liquids inside pipes, if you use the 1/10 capacity issue.

Using the small packet mechanic in pipes let's you pick pick the phase change temperature, but the 1.5C temperature rebound is still applied once the fluid is emitted into the world.

[0xFADE]

Side ejecting debris with the cooling setup avoids the bug because it is debris when it is combining with the other debris.

But I do have to add heat back to the system so it isn't as efficient.

[mathmanican]

27 minutes ago, ghkbrew said:

the 1.5C temperature rebound is still applied once the fluid is emitted into the world.

Have you checked this recently. When working on steam turbine heat deletion we used 1000g liquid water to get theoretical max (no 1.5C drop). I have not checked them all today but it would not surprise me if all pipes gasses and liquids avoid this.

30 minutes ago, ghkbrew said:

You mean the debris temperature reset bug

No. I mean the debris stacking bug.

33 minutes ago, ghkbrew said:

It's not about avoiding the hysteresis and changing exactly at the freeze/melt point. It's the fact that the 1.5C temperature rebound isn't applied. 

I see these as the same thing. It should require passing by 3C and then halving that in both directions.

Debris and piped liquids/gasses behave differently that tiles.

My suggestions for fixing this appear in the threads from last year. It is completely abusable and I'm excited to see what you do with it.

What we need is a simple, small chain of doubles. If you can figure out an easy way to harness the 2^6 energy gain, then the devs will be essentially forced to look at this. Fixing it could be as simple as making it behave as you thought it should. But beware as then materials with different SHC suddenly gain a new mechanic.

Another way to dodge the 1.5C phase change loss/gain is to use partial phase change mechanics (flaking).

[ghkbrew]

1 hour ago, mathmanican said:

Here is my most recent abomination on this topic, the aluminum phase change heat doubling experiment.

Very cool.  Is it energy positive yet? 

My one suggestion is to buffer the heat/cold sinks more for tighter temperature control.  I've got 1000kg of steam per tile in my build and no temp shift panels that can touch both sides of the reservoir.  They never get more than 1C away from the target temperature.

13 minutes ago, mathmanican said:

debris stacking bug.

I'm not familiar with that one.  Do you have a link?

[mathmanican]

1 minute ago, ghkbrew said:

Very cool.  Is it energy positive yet? 

Gain about 1370W for a cost of 800W. These are estimates. 

2 minutes ago, ghkbrew said:

buffer the heat/cold sinks

Only the cold sink leading to the turbine needs buffering. Liquids that drip out the door do not thermally interact with the door provided there is enough space to let the liquid spread out to prevent backflow conduction. 

4 minutes ago, ghkbrew said:

Do you have a link?

On my phone so a bit tough to search. Look for the fridge cooling by @Blazing Falken.

[Gurgel]

What a pity that this physical reality does not work like Oni Physics!. We would have global warming, energy, etc. all solved. Of course, the usual defectives would still insist on having war and poor people and sickness...

[mathmanican]

3 hours ago, ghkbrew said:

but the 1.5C temperature rebound is still applied once the fluid is emitted into the world.

The plot thickens.  Figured I would test some things. I chose the phase change temp to be at 380K (99999kg of hydrogen at 380C) for liquid. Upon phase change, the liquid actually dropped a full 3K, not 1.5K. I waited long enough for the room to essentially stabilize. 

  

We both have some more learning to do on this topic.  I am going to play with this some more to get all the details down, and return to @wachunga's experiment back with steam turbines. 

I changed the desired temp to 500, and sure enough, things stabilize in the steam room at 497. 

So phase changing liquid to gas inside pipes actually results in heat deletion of 3K. Fun. 

3 hours ago, ghkbrew said:

I'm not familiar with that one.  Do you have a link?

You probably found it, but this is the most up-to-date one that I know of. 

 

@wachunga, the debri stacking bug from @Blazing Falken, the right-to-left liquid flow bug, and the similar gas flow bug, all have to do with something retaining the incorrect mass for thermal computations. They all allow thermal amplification of whatever large mass's temp we want. Could they all essentially be the exact same bug, namely that weighted averages in thermal computations are using the former mass of something, until that something gets extra mass added to it.  They all seem like the same incorrect application of an average value formula. Could all three be swatted and squashed in fell swoop? @klei.ruby, if this helps spark any new insights into what you guys were trying to fix here, hooray! Gonna back link that page to here as well. 

Thoughts? 

[ghkbrew]

54 minutes ago, mathmanican said:

the liquid actually dropped a full 3K

I think this is exactly @wachunga's steam turbine phase change bug.

It seems that when a liquid is past its evaporation point, every tile that interacts with it thermally will apply the 1.5C temperature rebound independently.  This fits with my observation that a tile will not phase change unless it exchanges heat with something.

(Except debris for some odd reason.  A thermally isolated tile can be heated or cooled past it's phase change point by a debris pile without actually phasing. I have vague plans to exploit this for to superheat massive amounts of fluid without using 1kg/s pipes)

[TripleM999]

So use bridges to apply heat, for no phase change rebound, as far as i remember over all the experiments (also the ones in steam turbines, where insulation and bridges worked better than metal tiles). And i think, although not sure, debris melting on conveyor does not show a hysteresis or rebound. The ice on my debris conveyor thingy melts on the conveyor at exactly -0.7°C, and leaves a basket at this temp. This basket holds this temp even in regards to the conveyor temperature sensor, as i stumbled over in an early build.

[mathmanican]

4 minutes ago, ghkbrew said:

phase change bug.

Yep. That explains this paricular issue perfectly. I forgot that bug briefly. Thank you. Too many too keep in my human RAM. There were lots of others above as well.

6 minutes ago, TripleM999 said:

The ice on my debris conveyor thingy melts on the conveyor at exactly -0.7°C

Debris defintely don't have the 1.5C issue. Your observations match mine exactly.

I wonder if by cooling gas on 3 or 4 sides with liquids or solids, can we can generate more heat than we lose.

[ghkbrew]

13 minutes ago, TripleM999 said:

So use bridges to apply heat, for no phase change rebound

Unfortunately that doesn't work. Buildings (I tested temp shift panels and bridges) still cause a 1.5C rebound when they cause a phase change. But, interestingly that rebound doesn't stack.

1 building will cause a 1.5C rebound

2 buildings will cause a 1.5C rebound

1 building and 1 tile will cause a 1.5C rebound

1 building and 2 tiles will cause a 3C rebound

1 building and 3 tiles will cause a 4.5C rebound

[mathmanican]

2 minutes ago, ghkbrew said:

Unfortunately that doesn't work

Did you check if the buildings are hotter than the material?

[ghkbrew]

2 minutes ago, mathmanican said:

Did you check if the buildings are hotter than the material?

yeah, buildings (and tiles) will still cause evaporation if they're hotter than the water.  I tested 500.1K vs 500K water

But not if they're exactly the same temperature.  The requirement seems to be some heat exchange even if it's minuscule (cermainc insulated tiles).

[mathmanican]

10 minutes ago, ghkbrew said:

buildings (and tiles) will still cause evaporation

Even an automation wire does the trick. 

[ghkbrew]

28 minutes ago, mathmanican said:

I wonder if by cooling gas on 3 or 4 sides with liquids or solids, can we can generate more heat than we lose.

I appear to have broken something.  1 tick after painting 360K steam.

[mathmanican]

5 minutes ago, ghkbrew said:

I appear to have broken something.

That's pretty common (you can find people complain about it lots of places).  What happened, best guess, is that the steam transformed to liquid, which upon phase change enters drip form (not bead), and falls. Then the steam/water combo creates the problem. Seems to be that somewhere in the game code, whenever anything looses mass, there are bugs attached. Here the steam doesn't correctly get destroyed. In the debri stacking bug, the old mass gets used for heat computations. In liquid and flow, old masses get used.  Something is not correct in whatever algorithm runs the "lost mass" bit of code. 

Oh, wait a bit longer on that one, and you'll see the liquid actually returns (overriding the steam).  I just tried the same thing.

[ghkbrew]

Nevermind. It forms a droplet which takes a second to fall before transforming 361.5K water