(Pre-Patch) Decrypting Flaking (Partial melting/evaporation, Sweating, Flash Boiling, Flash Melting, Phase Fission)

[mathmanican]

Update: The mechanic is still in the game, but the equations have been updated.  I haven't had time to put together a new updated page on this, but you can find the correct mechanics located in the following post. 

Spoiler

 

Let's first pause and honor @Yothiel's work on decrypting heat transfer, and @wachunga's post on heat transfer equations, two excellent references for trying to understand exactly how heat flows in ONI. I'd also like to give a special shout out to @Zarquan who has repeatedly posted on this topic and been a huge help in tracking everything down, as well as @TripleM999 for actively engaging and contributing to the fun discussion over the last week as we tracked this down.

What is Flaking (Partial Melting/Evaporation)

ONI has a mechanic in place that causes solids and liquids to instantly phase up, and "flake" off, small 5kg bits of matter, ignoring thermal conductivity. This allows a large tile of ice to slowly "sweat" liquid without having to change all at once. It's been in the game for a long time (maybe since the beginning), and discussed in many places under the names of sweating, flaking, flash boiling, flash melting, melting, burping, and more. I collected a lot of those links here (page 4 comment about half way down). I have decided to settle on the term "flaking" as it appears in the dev update notes, when they removed the ability to flake man-made solid tiles. We could call this "phase up fission", as we are splitting a cell into two cells, with one of them phasing up. Update: The devs have named it "partial melting" and "partial evaporation" here. 

Until recently, the exact mechanics under which this occurs have not been on the forums. @wachunga empirically derived two formulas, and I've spent the last week exploring their consequences, tracking down conditions and limitations (and of course massively abusing it). This post will share the conditions to enable flaking, leaving abuses, exploits, monstrosities, etc., to other posts.  Update: Here is an introductory post showing 4 key examples. 

Key Terms (Donor, Parent, Child, PhaseUpDelta)

To give the formulas, we need to first define some variables:

• Donor - The hot cell that gives away some heat to enable flaking. Other possible names could be catalyst, or enabler. I like @wachunga's choice of "donor" as the cell donates heat to enable flaking. 
• Parent - The cell to be flaked. This cell will loose mass when flaked. 
• Child - The new chunk of phased up matter, which always appears in 5kg chunks at exactly 3C above the phase up temp of the Parent (so melting point if the parent is solid, or vaporization point if the Parent is a liquid - though I'm not sure I'd call crude oil's phase up point a vaporization point). The child is the offspring of the parent.  
• PhaseUpDelta - The difference between the Parent's current temp (after conduction has been computed) and the Child's temp (3C over the Parent's phase up temp).

Let's look at 2 examples to clarify these definitions. 

• A block of 1000kg ice at -10.6C comes in contact with 20kg of hydrogen at 10C. The ice will be flaked to produce 5kg of water. The Donor is the 20kg of hydrogen. The Parent is the 1000kg of ice. The child is the water, which will appear at precisely 2.4C (-0.6C+3), and the PhaseUpDelta is the difference between -10.6C and 2.4C. Note that there is a very clear relationship between Parent and Child, namely the Child will always be a phased-up version of the Parent. 
• A 10kg blob of crude oil at 200C falls in bead form and brushes past a hot igneous tile kept at 450C. The Donor is the igneous rock. The Parent is the crude oil. The Child is a 5kg blob of petroleum that will appear at 402.85C (399.85+3). The PhaseUpDelta is the difference between 200C and 402.85C (It's actually a tad less than this as the hot igneous rock will transfer a bit of heat to the crude via conduction first).  

Spoiler

The Parent-Child words lets us think about creating a child from a parent.  We could go with Block-Chip and then take "a chip off the old block." Or we could use "Scalp-Flake" if we want to constantly think about dandruff... I'll stick with Parent-Child as the English idiom and may not translate well, and I'd rather not think about dandruff.

Conditions for Flaking to Occur

The mechanics, both formula and conditions, depend on whether the Parent is a liquid or a solid. 

Flaking Liquids (Liquid Parent)
When the parent is a liquid, the key equation (units in kg, and C or K) is 

DonorNewTemp = DonorOldTemp - 10 + (5 * ParentSHC * PhaseUpDelta) / (DonorSHC * DonorMass).

The conditions for flaking to occur are as follows:

1. The Donor can be solid, liquid, or gas. 
2. The Donor must be adjacent to the parent after the game computes new tile locations, but before it updates their location. As such, you can flake liquid from 2 tiles away, with the 5kg child constantly between the Donor and Parent, never allowing the Donor and Parent to exchange heat via conduction.  
3. The Parent must have a minimum mass of exactly 5010g (so 5009.99999g fails). 
4. The Donor's new temp must be below the Donor's old temp using the formula above (within some small error). Heat conduction is probably updated before this condition is checked (so use the DonorOldTemp value that would result from regular conduction when you compute the PhaseUpDelta). 
5. The Parent liquid cannot be trapped (it must be able to be displaced).  
6. The Parent liquid cannot be too close to its phase up temp (I think within 3K), and the donor must be 3K over the phase up temp. So there will be at least a 6C difference. 

If all the conditions above are met, then a 5kg Child, at exactly 3C above the Parent's vaporization (phase-up) temp, will appear exactly where the parent was. The Parent is displaced, sometimes in quite interesting ways, similar to the way high mass beads expand. If the parent is trapped (cannot be displaced), then liquid flaking does not occur. For example, venting crude oil through a single vent on the bottom tile of a lake of petro will not enable flaking as the crude is trapped by petro. However, venting crude oil on two adjacent tiles at the bottom of a petro pool will enable flaking, provided the tile under them is hot enough.

Flaking Solids (Solid Parent)
When the Parent is a solid, the key equation (units in kg, and C or K) is

DonorNewTemp = DonorCondensation + 6 + ((5 * ParentSHC * PhaseUpDelta) / (DonorSHC * DonorMass)).

The conditions for flaking to occur are as follows:

1. The Donor must be a gas (I tried both liquids and solids to no avail). 
2. The Parent must have a mass above 5kg (so 5000.001kg works, but 5.0 does not).
3. The Donor's new temp cannot exceed the Donor's old temp - that's it (I checked this down to 0.1C). Note that to compute the PhaseUpDelta above, you must use the Donor's Old Temp that is computed AFTER regular conduction has already occurred, which means if you have a really large temp difference between Donor and Parent, then it may be hard to spot this formula (the increase in Parent temp due to conduction with the Donor decreases the PhaseUpDelta). @TripleM999, there is no "cannot phase change" condition here.  Though note that if you manage to get a large enough blob of gaseous gold to flake off something, then the gold's temp will be dang close to the condensation point and swap to liquid in the next tick (ending flaking for solids, as it requires gas).  It is possible to flake regolith to get a 5kg/tick magma flow (so 25kg/s). 
4. The Parent solid cannot be too close to its phase up temp (I think within 3K), and the donor must be 3K over the phase up temp. So there will be at least a 6C difference.  

The child appears in drip form (not bead form). It appears on the same side of the parent where the donor is located (top/bottom/left/right), though for left/right it seems to appear diagonally down on that side, if no solid tile is in that spot (allowing you to keep a donor cell immediately left or right, with the Child liquid draining away under the Donor). If there is room for the child to appear (by either falling or displacing the Donor), then the child will eventually materialize as a 5kg chunk of liquid at the Parent's phase up temp plus 3C. Otherwise, the Child deletes itself and 5kg from the gas (so you can loose 10kg of mass this way, 5kg from the Donor and the 5kg child). If the Donor gas has less than 5kg, the Child mass drops by the Donor amount, and then the Child appears at the reduced mass in the Donor's tile. 

These conditions appear to fully describe if/when flaking occurs.  If you find any errors, please share. I have double and triple checked most things above. If you want to avoid flaking, just prevent a condition from occurring. If a build of your breaks, and you can't explain it, sometimes flaking is the issue. 

Pictures? Spreadsheet? More Examples?

I'll be putting up more posts with examples to abuse and/or prevent flaking.  This post contains the mechanics, in all their gory detail. Monstrosities will come soon. Feel free to add pictures of times flaking hit your build and ruined it (or make a post and link to it from here). Happy Flaking all. 

Update: Here is the first post, an introduction using examples which also contains a spreadsheet you can use. 

Edit: I added the within 3K condition to the bottom of each, as I remember that this was an issue, but have not yet double and triple checked it). In addition, I believe the reset temp on liquid flaking must not phase change the material. 

[Yoma_Nosme]

Thanks!

Great summary of the previous posts! I've read them with great interest but was unable to contribute anything constructive

But i have thought about a bit... in terms of fixing this bug...please correct me if I'm wrong... the bigger the donor mass the lower the donor new temp will be (edit: compared to same scenario with lower mass)...so wouldn't it be possible to cap max tile mass to mitigate the bug...so even when it happens it happens a couple of times and then the donor new temp would be higher than old and thus violating rule 4 for fluids or rule 3 for solids

[mathmanican]

1 minute ago, Yoma_Nosme said:

the bigger the donor mass the lower the donor new temp will be...so wouldn't it be possible to cap max tile mass to mitigate the bug.

This would dampen solid flaking's potential (but 20kg of hydrogen can do a lot of cooling still). As for liquid flaking, I can use 19kg of Naptha in 6 spots while flaking phosphorous to enable more than 18 steam turbines to run pretty much nonstop with less than one steel aquatuner providing the heat source (see here). Restricting large masses will do nothing towards fixing liquid flaking and heat generation abuses.  What it will do is prevent flaking from enabling massive cooling gains (from resetting hydrogen's temp down to 21K with 10000kg of gas - good luck heating that up). 

My best guess is that they wanted a way to simulate ice melting in the ice biome, with water dripping periodically.  It works, and the formulas they choose seem very balanced under those conditions.  Change the conditions though, and ..... um......  well .... I'm not sure I have the right words for this.

The key abuses in flaking liquids come from equating DonorOldTemp with DonorNewTemp, and then making the conditions as near this as possible on DonorMass and PhaseUPDelta. Sometimes you fix the DonorMass and locate the right PhaseUpDelta. If you have control over the mass, then you pick the PhaseUpDelta (Magma, right before it solidifies all the way up to RockGas), and then adjust the mass.  Building natural tiles, or using liquids/gases at the mass you want, are then your tools. 

For flaking solids, increasing the mass a ton just gets you the ability to reset the temp of the Donor to near its condensation point (think Hydrogen). Yeah, you stop flaking from happning on the next tick, but you also gained yourself a fridge that only gets more powerful the more gas you use.... To flake solids and melt stuff repeatedly, it is much more tricky, and small masses are what you want.  It can be done (melt 1000kg ice tile with less energy than normal), but is much tricker to abuse. 

[TripleM999]

1 hour ago, mathmanican said:

My best guess is that they wanted a way to simulate ice melting in the ice biome, with water dripping periodically.  It works, and the formulas they choose seem very balanced under those conditions.

I think, similar is true for oil biome and abyssalite (maybe planned for water/steam too). They wanted to make for some surprise, when crude oil hits hot abyssalite even with its marginal conduction, but it calms rather fast, with abyssalites final temp at around 300-400°C, and some left over petroleum and sour gas bubbles.

[MrGuga]

Could this explain heat deletion bug inside steam chambers? maybe because liquid is near phase change and something gets flaked... I just can't figure how... yet.

 

[mathmanican]

Just now, MrGuga said:

Could this explain heat deletion bug inside steam chambers?

If you only have one chamber, then you never reach 5kg of water. I thought the same at one point, but @Zarquan helped me see and I'm pretty convinced it's not the issue. Then I reverse engineered a contraption to gain heat from flaking on purpose in a steam chamber, but that lead to me discovering flaking mostly likely cannot happen if the temp of the material gets too close (withing 3K) to the phase change temp. Tricky stuff (and I had to use 7kg of Naptha - powerful stuff for flaking). 

I just spent the last hour watching gas flow to see what I could do about heat loss in turbines (as I've got a new cooker I'm making).  My current guess is it's a gas flow mechanic acting horizontally when you have unequal masses.  Just got some ideas to test...

[wachunga]

The big problem with figuring out the horizontal gas bug is that the mechanic likely responsible happens randomly. Thorough testing is tedious and unpleasant.

For anyone who doesn't know, adjacent gases of the same material will randomly engage in a temperature transference of a different nature than the normal conduction. Hot gas below cold gas will sometimes swap which creates a pseudo "heat rises" effect. Horizontal gases of the same (similar?) mass will average out their temperatures. Different masses and they swap temps instead. So high mass hot gas swapping temps with low mass cold gas deletes heat. Vice versa creates heat.

[TripleM999]

12 minutes ago, wachunga said:

Hot gas below cold gas will sometimes swap which creates a pseudo "heat rises" effect. Horizontal gases of the same (similar?) mass will average out their temperatures. Different masses and they swap temps instead. So high mass hot gas swapping temps with low mass cold gas deletes heat. Vice versa creates heat.

That is pretty much, what i observed in my NG reactor designs. Horizontal heat exchanger are much better for cooling, vertical for heating, cause of these.

Look at the temp overlay here:

 

[wachunga]

@mathmanicanVery minor bookkeeping suggestion, move the "-10" from liquid flaking to immediately after "DonorOldTemp". I initially viewed the temp and the SHCDeltaMass bit as variables with the -10 as a constant and I tend to arrange things as variables first with constants at the end. After doing solid flaking, in my mind the "+6" became less a constant and more a modification to the temp variable so I grouped it that way. In any event, my OCD is a bit triggered that the two equations aren't arranged the same way (especially so since I did it!) and I would advocate that they should be, however that happens to be. Making it more clear that the SHCDeltaMass bit is the same in both cases.

 

For the record, I like the parent and child terminology.

[psusi]

21 hours ago, mathmanican said:

I have decided to settle on the term "flaking" as it appears in the dev update notes, when they removed the ability to flake man-made solid tiles.

Ohh, I missed that one.  So bunker tiles will no longer flake from hot rocket exhaust?

21 hours ago, mathmanican said:

When the parent is a liquid, the key equation (units in kg, and C or K) is 

 

Shouldn't the units be grams since that is what SHC is normally listed in?

21 hours ago, mathmanican said:

DonorNewTemp = DonorOldTemp + (5*ParentSHC*PhaseUpDelta)/(DonorSHC*DonorMass)-10.

Weird... so if the phase up delta and/or SHC of the child are small enough, the donor can actually lose temperature?

21 hours ago, mathmanican said:

DonorNewTemp = DonorCondensation + 6 + ((5 * ParentSHC * PhaseUpDelta)/(DonorSHC * DonorMass)).

What is DonorCondensation?  Shouldn't that be DonorOldTemp?  And so solids always gains at least 6 degrees?

22 hours ago, mathmanican said:

The Donor's new temp cannot exceed the Donor's old temp

Now I'm confused.  The equation seems to indicate that the donor's temp should always go up.

[mathmanican]

40 minutes ago, wachunga said:

move the "-10" from liquid flaking to immediately after "DonorOldTemp"

Done. I also added spaces to help with maintaining a consistent style (I generally just let LaTeX do this for me). 

40 minutes ago, wachunga said:

I like the parent and child terminology.

Thank you.  I agonized over this for way too long (i have a huge list of other options). 

29 minutes ago, psusi said:

So bunker tiles will no longer flake from hot rocket exhaust

Correct.  They dealt with this issue.  If you get them to flake again, please post it somewhere.

I think the devs know exactly what is going on, and could change it in an instant.  Whether they want to or not is another story.  I love this game, and love Klei's work. If they decided to leave flaking in the game, as is, I'll survive and just add it to the list of ONI physics rules to play with.  I don't know if they'll be OK with the monstrosities that follow from the consequences of flaking, so I suspect we'll see a change someday. We just gotta provide more and more absurd monstrosities, till someone at Klei says, "This has to stop!"  I do hope they laugh at little, not get stressed, when they see these physics abusing monstrosities.

The joy Klei has provided me in letting me play with this world is worth tons more than the cost of the game. A huge thanks to everyone at Klei.

32 minutes ago, psusi said:

Shouldn't the units be grams since that is what SHC is normally listed in?

The units in the equation are not balanced. I doubt the devs were thinking, "let's keep the units" balanced when they did stuff.  We could add 1000 into the equation somewhere to fix this, but I'll leave it for now. 

34 minutes ago, psusi said:

Weird... so if the phase up delta and/or SHC of the child are small enough, the donor can actually lose temperature?

The donor always drops in temperature when flaking occurs.  The Donor donates heat to allow the Parent to create a child. If the value of the donor's new temp were to exceed the old temp, then no donation is made, and the child cannot be created. 

36 minutes ago, psusi said:

What is DonorCondensation?  Shouldn't that be DonorOldTemp?  And so solids always gains at least 6 degrees?

In solid flaking, the Donor is a gas, not the solid. This equation is definitely an odd one, but is correct as written.  Donor condensation is the temp at which the gas would condense to liquid (so for hydrogen this is -252.85C). The solid temp does not change at all from flaking (but it does from conduction).  It is the Donor's temp that is reset. That's what makes this truly crazy.  Try, in sandbox, painting 1000kg of hydrogen gas (at 100C) next to 1000kg of ice (at -40C).  The results will amaze you, and show you just how abusable this mechanic is. Cooling has never been easier (though you might freeze your entire asteroid if you get too much hydrogen).  Even an army of  liquid tepidizers can't keep up with the sheer heat deletion that is possible. 

41 minutes ago, psusi said:

The equation seems to indicate that the donor's temp should always go up.

We moved the -10 to the front of the liquid parent equation to help with this.  For the solid parent equation, the fact that we use Donor Condensation temp should help see this is not true.  It also helps see why you need at least a 6K difference between Parent and Donor.  To abuse a hydrogen ice combo, you do have to maintain a 6K differential, but that is very easy, especially since the instant drop in hydrogen temp will drop the ice temp. 

[beowulf2010]

2 hours ago, mathmanican said:

 I do hope they laugh at little, not get stressed, when they see these physics abusing monstrosities.

I know I do every time you, Neotuck, Saturnus, and the others make something ridiculous. 

I still cackle remembering your gas converter from a while back. 

[Tobruk]

You have quite a talent of explaining the more difficult ONI concepts. I've read the previous posts on flaking and even after a couple of them I haven't had a clue what's going on. Your post cleared up any doubts. The definition and the two examples definitely helped.

[Prince Mandor]

Someone experienced check my thoughts please.

Lets take some amount of steam. 4.7 kg/tile. Put it near abyssalite and start heating it (for example, with diamond tile and liquid steel flowing)

So, new temperature of steam after flaking must be

DonorCondensation + 6 + ((5 * ParentSHC * PhaseUpDelta)/(DonorSHC * DonorMass)).

96.35 + 6 + ((5 * 4 * (3421-abyssalite temp)/(4.179*4.7))

about 3585C for 0C Abyssalite

So, if steam heats up to 3'585, then condition (Old temp > New temp) became fulfilled and it just flake entire (except 5kg) abyssalite tile into tungsten?

Is it really so simple?

[mathmanican]

3 hours ago, Prince Mandor said:

Is it really so simple?

Just checked in debug what happens. A couple things to note.

1. One the temp resets, then the condition is no longer true, so you do have to add a tiny bit of extra heat. 
2. Abysalite does conduct heat with the gas (a tiny amount), but that amount does reduce the PhaseUpDelta.
3. If you don't have a place for the tungsten to fall, then the mass of the gas greatly increases and you loose a ton of heat. 

So, as long as you design arounnd those limitations, it should be that simple. Please share when you've got a design.

[wachunga]

In addition to what @mathmanicansaid, DonorCondensation is 99.35 and PhaseUpDelta is 3424.85 for 0C abyssalite. The extra 3C gets added onto the parent melting point. That gives a donor temp of 3592.746455C. In game you are limited to 3 decimals when painting, but that is indeed the critical temp. Above that flaking occurs, below that no flaking. After each flaking event, add another 0.001C worth of heat to the steam and it flakes the abyssalite again, creating 5kg of 3424.85C tungsten each time. Massively broken.

[mathmanican]

4 hours ago, Prince Mandor said:

check my thoughts please

Here is a screenshot of all the computations. I really should just get a googlesheet set up that does this for people.

Thanks @wachunga for mentioning the other two bits. I just looked at the idea, quickly built a contraption (vertical tile of steam to the side in which fresh tungsten could fall), and then started raining tungsten.

[Manarz]

Okay so we are not finished here...
While testing for the turbine heat deletion thread i wanted to check wether high steam pressure would flake the incomming 95C° water.
Surprise: it kinda did but not rly...

According to this formula
DonorNewTemp = DonorOldTemp - 10 + (5 * ParentSHC * PhaseUpDelta) / (DonorSHC * DonorMass).

if id have really high pressure in the steamchamber i would expect massive heatloss if flaking would occur.
take for example 9999kg steam @120 C° with 3 steamturbine outputs comming in (6kg Water). This should result in 10°C heatloss in the Donor steam cell.

So here is my test and its result:

results:

1. Picture: 1 Frame before flaking
2. & 3. Picture: Frame where flaking happend
3. Picture: 1 Frame after flaking without any visable heatloss
 

Is it flaking? If so does Gas Liquid Parent - Gas Donor flaking behave diffrently?
Or is it a thing which happens when u use the same Material for Donor and Parent and is just vaguely related to flaking?

 

9 hours ago, mathmanican said:

Here is a screenshot of all the computations. I really should just get a googlesheet set up that does this for people.

 

id use that.
 

[Zarquan]

1 hour ago, Manarz said:

Is it flaking? If so does Gas Parent - Gas Donor flaking behave diffrently?
Or is it a thing which happens when u use the same Material for Donor and Parent and is just vaguely related to flaking?

This isn't gas parent.  The parent is the liquid water, the child is the steam. 

This most certainly is flaking. 

That brings up an interesting point.  If we carefully calibrated the steam mass, could we construct a simpler self powered aquatuner by simply using water to flake steam?  Or perhaps use something heated by the steam to flake water?

[mathmanican]

1 hour ago, Manarz said:

1. Picture: 1 Frame before flaking
2. & 3. Picture: Frame where flaking happend
3. Picture: 1 Frame after flaking without any visable heatloss

The temp of the parent has to be below 3K of the vaporization point AFTER conduction is calculated. That is why you didn't see flaking. Conduction won. That's my best guess.

1 hour ago, Manarz said:

id use that.

I'll put together something. I was gonna wait for an abuse thread, so maybe I'll add bells and whistles on the abuse thread sheet.

20 minutes ago, Zarquan said:

This most certainly is flaking. 

On second pass it IS flaking. Maybe the condition for liquid flaking is checked before conduction. This was the thing I have not triple checked. It definitely was flaking. 5kg of steam at 102.4 is absolutely it. Notice how the parent liquid got displaced.

I spent way too long checking solid flaking condition on when conduction occurs that I didn't have the desire to check liquid flaking.... maybe it's time. I already built a turbine design in the part 2 thread that used water flaking in a turbine design. I just need to revisit it now with current knowledge...  Too much to do.

[Manarz]

54 minutes ago, Zarquan said:

This isn't gas parent.  The parent is the liquid water, the child is the steam. 

my bad! its getting late :S i had the right calculations in mind though.

 

55 minutes ago, Zarquan said:

That brings up an interesting point.  If we carefully calibrated the steam mass, could we construct a simpler self powered aquatuner by simply using water to flake steam?  Or perhaps use something heated by the steam to flake water?

that was exactly what i had in mind originally. but the formulas didnt check out so i posted my findings for u guys.

[mathmanican]

18 hours ago, Prince Mandor said:

Is it really so simple?

Use the fact that you can embed abysallite in abysalite tiles and you can convert all abysallite on map quite quickly. See here from @kbn. 

Spoiler

Image from @kbn.

You can also use this idea to rapidly liquefy any substance, at basically any volume. Here is a link to the thread. 

Spoiler

 

 

[mathmanican]

Perhaps the simplest way to explain flaking is to watch the adipose episode of Doctor Who. Except in that episode the child is exactly 1 kg instead of 5 kg.

Spoiler

 

 

[bobe17]

can liquid flaking occurs with built tiles ? i have some strange behavior with liquid oxygen and an insulated tile made of insulation. 5kg oxygen goes back in gas form at -180°C and the temperature of the tile drops by -10K (-189.7°C now and stable; the neighboring tiles start doing the same thing). This looks like flaking.

[Prince Mandor]

1 hour ago, bobe17 said:

can liquid flaking occurs with built tiles ? i have some strange behavior with liquid oxygen and an insulated tile made of insulation. 5kg oxygen goes back in gas form at -180°C and the temperature of the tile drops by -10K (-189.7°C now and stable; the neighboring tiles start doing the same thing). This looks like flaking.

flaking occurs with natural tiles of liquid oxygen. They are flaking.

And yes, built tile can be temperature donor, as far as I know