Gas Heat Transfer seems completely broken ?

[Lifegrow]

Hey folks, hoping for some help/understanding.

I'm back playing ONI after a good ol' break and trying to get my head around some of the many changes - but I can't understand for the life of me what has happened to gas temperature dispersal/balance/transferal - whatever the technical term - the heat transference between gas a) and gas b) (and I think c) in this case)

What the heck has happened to packets of gas released into a gassy atmosphere, and the janky way they exchange temperature? I'm hoping some of you who have been around amidst all the odd choices that have been made lately (don't get me started on lavender-coloured pipes...) can help explain what's going on here.

The thing that caught my attention was this large gas-filled debug room. I'd been doing some tinkering off screen, and have been letting a build run for ~25 cycles. I hadn't touched this room, except to do some tinkering with some sweepers/loaders and then left it whilst I was messing with something else. Why isn't the temperature averaging out?

I'm aware that tempshift plates exist, but since when did calculations stop being calculated within this simulation? The below 2 pictures show the high/low temperatures. The bottom third of the room is a static -15.5°C, the middle third ranges from -15.5°C - 1°C, then the upper third from 1°C - 68.8°C. Again, this is after 25 cycles.

What have I come back to?   I understand that heat rises, but this seems very broken to me. At this point i'm 3,700 hours deep into ONI, and I've not seen worse temperature calculations since heat fell like a waterfall way back in the agricultural upgrade... 

Can anyone explain this? There's a bunch of pictures below, if anyone can shed any light i'd appreciate it.

Cheers,

-Life
 

Spoiler

 

 

I then decided to do a couple of quick vertical tests just to have a play about (4 chambers, all filled with 20kg of 0°C Hydrogen, then inserting a single 20kg packet of 400°C hydrogen into the lower left/upper right to see the difference.

 

Spoiler

 

Start:

 

After 10 Seconds :

After 30 seconds :


After 1 minute (note that in the single file tube on the right, the hydrogen packet 6 tiles below the 400°C packet has yet to change) :

 

After 3 minutes : (at this point, i'm bored)

 


 

 

 

[mathmanican]

Something weird happened in the fix to temp swapping.....

[TripleM999]

That hot gas at the top takes eons to exchange heat with colder bottom, was already there before the temperature swap fix.

I think, before the fix it was not that obvious as it is now. Gas to gas heat transfer is probably much too low in comparison to gas to other mediums. And the "swap hot gas up" mechanic makes things worse.

[Smife]

Even the temp overlay looks weird now, but maybe it's just me.

[OxCD]

Dammit, and it's hydrogen. Couldn't imagine with CO2 or Cl...

It seems there's a decision to make on devs' side

[KittenIsAGeek]

Gas temperature has always preferred to transfer heat up rather than down.  I've used that mechanic in the past to protect my farms from heat producing buildings before I could get active cooling into place.  I haven't noticed any anomalies with my base over the temp-swap bug.  Below, I've got cooling at the top, heating at the bottom, and a smooth transition between them.

Spoiler

Here's a room filled with CO2.  It was on average 45c when I started cooling.  My coolant line runs behind the petrol generators.  The bottom is still warm simply due to the mass of hot water that got compressed there.

Spoiler

I remember when thermal expansion first hit.  The tube on the left of @Lifegrow's test is how gas thermal transfer originally behaved and I made a lot of my models based on it.

[nakomaru]

This is pretty neat. I like how up-swapping can result in layers, and it seems in line with the heat transfer equations we've always had, but perhaps gas to gas transfer could use a little bonus to heat transfer. Even without it, is this impeding some gameplay or designs? Kitten shows how to design with it in mind.

In the 25 cycle example, it seems there's a substantial mass (probably much more than the gas below) of warm material up near the top. I wonder how it would look if built on the bottom.

[KittenIsAGeek]

I was looking for some old screenshots and I haven't found them (yet).  Anyway, when the Thermal expansion hit, there was nearly no side-to-side transfer of heat through gas.  You could, for example, have a generator 3 tiles to the side of a farm, and your farm would never overheat because the heat would only go up from the generator.

I suspect the temp-swapping stuff was added as a fix to this phenomena, helping the heat to diffuse out and adding extra bugs in the process.

Hah! I found one.

Spoiler

This was documenting a bug where a puff of polluted oxygen (or other similarly small packet of matter) would gain heat energy from a building (say, a battery) regardless of the mass involved.  Fun times.  Anyway, the point for THIS post is that as you can see, the heat from the generator is going up and not really spreading out much.

[TripleM999]

47 minutes ago, nakomaru said:

but perhaps gas to gas transfer could use a little bonus to heat transfer

 

9 minutes ago, KittenIsAGeek said:

I suspect the temp-swapping stuff was added as a fix to this phenomena

At least the horizontal heat equalizing is yet in, the temp swap was there for handling the corner case, when equalizing would push the temperatures of one tile beyond origin temperatures. So there is some boost to conduction within homogeneous gases, when these tiles are not to far away in pressure, but only horizontally.

[Lifegrow]

49 minutes ago, nakomaru said:

Even without it, is this impeding some gameplay or designs?

I think the calculations being so slow over the vertical plane is worrying. Impeding isn't the word i'd use at this point, although for players who don't frequent the forums/reddit, it'd probably be confusing. As I said initially i'm aware that tempshifts exist but they shouldn't be the solution to the problem shown in the vertical tests. Nor should we be looking for solutions to problems that have been created by bodged changes made to the game hastily and without proper testing in my humbly-bearded opinion.
 

57 minutes ago, nakomaru said:

In the 25 cycle example, it seems there's a substantial mass (probably much more than the gas below) of warm material up near the top. I wonder how it would look if built on the bottom.

Yeah, lots of debris  - I was working on a storage solution so needed "crap" to sort  If it helps, there was just as much on the floor of the room if you look carefully.

20 minutes ago, KittenIsAGeek said:

You could, for example, have a generator 3 tiles to the side of a farm, and your farm would never overheat because the heat would only go up from the generator.

This is what I called the "heat waterfall" - probably bad wording, but when cold/hot would rise/fall in very exaggerated ways. This was changed a good while back, and was the main culprit of the surface cooling (the dreaded drip cooling) before it was changed. Things were much better for a long time - this has been changed since I last played in Dec 2019 (so within the last 6 months).


As far as how damaging this is, I really have no idea of the full implications yet - it's only when we find more builds-behaving-badly, and explore the broken things in detail that we finally start to understand what causes x to happen. I was recently toying with a build using liquid CO2 to condense steam - not a tricky concept, and one we've no doubt all done before at some point. This time however it was definitely impacted by this change, and in the specific build I was working on, I ended up needing multiple radiators/tempshifts/bridges and unnecessary malarkey to simply average out the temperature of gas within a 20x10 room. That isn't the ONI I left 6 months ago, something has been fundamentally broken.

[mathmanican]

3 minutes ago, Lifegrow said:

explore the broken things

Before the update a steam turbine would stabilize in gained wattage fairly quickly. Now the wattage jumps all over the place because the steam underneath does not equalize in temp, ever. Yes, something changed. 

[Lifegrow]

2 minutes ago, mathmanican said:

Before the update a steam turbine would stabilize in gained wattage fairly quickly. Now the wattage jumps all over the place because the steam underneath does not equalize in temp, ever. Yes, something changed. 

My issue here is I wasn't around to know at what point this was broken (and i've got a 6 month gap) else i'd have damned well pointed it out in the patch release thread   

Are you saying it may have been in the latest few patches?

[mathmanican]

20 minutes ago, Lifegrow said:

Are you saying it may have been in the latest few patches

It was the last update. Period.

@Lifegrow check here. I did mention the weird wattage issue. 

 

[Lifegrow]

@klei.ruby Anything known at your end about this? Was this the intended mechanic going forward ?

[cpy]

Completely normal phenomenon!

[Lilscratchy]

This is making me scared to chuck in any build involving any kind of heating/cooling or heat transfer in order to work. 

Is it really that bad? Ive been tinkering with some designs yesterday and have seen more time in the temperature overlay than i'd wish on my worst enemies but i haven't really noticed anything horrifyingly off.

Only thing i noticed is that when i dripped packets of 10kg liquid oxygen into a roughly 80°C oil biome it seemed to not really equalize temperatures with the hot CO2 in there as quickly as im used to. 

[TheMule]

I've noticed the same. I'm experimenting in ways of making use of pwater (something I didn't care much in the past). For example, I've always assumed that the "best" (w/o involving sour gas, that is) way of using oil reservoirs was the loop: 95C water  -> wells -> boiler -> generators in a sauna -> 95C  water. But with that I'm left with other 95C sources of water... (salt)water geysers, CSV, SV, all produce 95C water.

In recent colonies I'm experimenting with keeping the generators around 40C and having 40C polluted water as output.

Long story short, I'm playing with arbor trees (domesticate) and ethanol distillers... I have about 10 of them 4 tiles above the trees. I've noticed all the tree tops stifled for high temp (of course, the distillers are hot), but to my surprise, heat wasn't affecting the other branches...

Turns out, I have a single layer of CO2 (at high pressure) quite hot but the layers below are 25C.

There's basicly a natural vertical insulation in gas masses, even with 6kg/tile of CO2 (granted, CO2 isn't the greatest conductor but still).

 

Edit: I can't think of way to abuse of this, or even builds being affected. Vertical transfer of heat was never huge, basing a build on that was never a good idea. It's just now it seems too close to perfect insulation...

[TheMule]

Edit2:

I'm cross linking this, 

a more extreme example of it. That's what I mean, it's near perfect vertical insulation.

 

[mathmanican]

Well, I decided to rewrite the post that I have hopefully since hid (it was wrong).  

I painted 1kg of CO2 at 300K next to 1kg of CO2 at 310K.  I did this both vertically and horizontally.  I put the hotter gas on top to avoid a vertical temp swap.

Here are the temps I recorded, and they match both vertically and horizontally. 

Below I have placed the predicted temps, using the conduction equations that we've deciphered and someone collected on the wiki. 

Heat conduction is working exactly as we would predict it to.  Yes, it seems a little slower than normal because some of the temp equalizing algorithms have been altered. 

Perhaps the gas - gas interaction need a new multiplier? Maybe it is fine as is.  @Lifegrow's comment to start this thread suggests that he is not very happy with how slow gas heat transfer is now.  I definitely think this is worth looking into, but right now I do not see any evidence to suggest that gas transfer is bugged.  The only thing that occurred was a removal of the equalization algorithm which lead to heat deletion. 

[Lifegrow]

Excellent work as ever bud, can't see the pics just now but I trust you wholeheartedly.

5 hours ago, mathmanican said:

Yes, it seems a little slower than normal because some of the temp equalizing algorithms have been altered. 

This is the issue here - the speed.

It's changed a lot, and I've since done further testing and the rate it exchanges at is terrible. It's often difficult to see within builds that are already toasty-red-hot, and they're the builds that would be affected the most. I remade an old build where I bubbled liquid methane through sour gas to aid in pre-cooling, and it runs like a two legged dog now (the rear legs are missing, it's dragging its behind...) I used this build because it showcases extreme temperatures of ~ -163 liquid methane boiling, changing state, then bubbling through sour gas.

I think there needs to be a generous tweak, but sadly I fear Klei are currently in the business of making changes without thorough testing. Few of the devs seem all that bothered about late game builds (let's face it, late-game mega bases were never their main concern) and so the knock on effects of changes like this don't seem to resonate all that much. I honestly doubt it was even tested/compared before they rolled it out the change.

Edit - as an afterthought - any chance you can further add to those experiments with a taller/longer chamber - maybe 7x1 ?

[mathmanican]

2 minutes ago, Lifegrow said:

doubt it was even tested/compared before they rolled it out

They did mention, in the bug file, that they were doing internal testing, and then only rolled it out a few days later. Of course, the testing AFAIK was only to make sure it did what they hoped it would, prevent the heat deletion issue.  What was probably not tested (nor maybe even considered) was the larger effect on heat transfer speed, in general.  

Gas-gas transfers at the same speed as solid-solid, with one difference.  Hot gas will get swapped upwards, resulting in very interesting layers of heat. More than ever before, the vertical swapping of gas will yield a "Hot air rises" effect in the game.  So that's a good thing. 

FYI, even before the change happened, you would get the same layering effect ( @TripleM999's test showed this).  However, once a single blob on one layer got cooled down, that cooling could spread horizontally quite rapidly (from the equalizing effect that got axed).

The fact that gasses now exchange heat more rapidly with tiles than they do other gasses is definitely a little off. I don't think most people will think, "To even out the temperature of these gasses, I should build a ceiling between them." That is precisely what the current state yields. If you built a horizontal row of tiles (metal or igneous) with a gap every other tile, then the tiles would increase the spread by a factor of 25.  Not intuitive, as we generally build horizontal walls to separate floors (and likewise slow gas conduction). 

18 minutes ago, Lifegrow said:

Edit - as an afterthought - any chance you can further add to those experiments with a taller/longer chamber - maybe 7x1 ?

Sure. Probably tomorrow though.  If you want to do the same, and share, please post. 

[Lifegrow]

Woah now, let's stick to what we know. If you want advice on beard products, pizza toppings or a collection of expletives to express the mood - I'm your man. When it comes to excel, maths, and god-tier nerdgasms, i'm looking your way bud  

(love you. xox)

[Lifegrow]

Ok : the plot thickens - little video uploading now, @mathmanican @TripleM999 you may want to see this... 
 

 

[TripleM999]

Now i see what they have done.

Heat Equalizing.sav

They completely removed the equalizing, not only the buggy temperature swap, but the heat neutral quick equalizing too. The fields, i painted in the save, would have the instant equalizing before the patch. Now they do something completely different, a complete tile swap, similar to the top-down-swap. This explains the more violent fluctuating gas behavior. With equalizing completely vanished, it exhibits the minuscule conductivity of gases more prominently.

[Lifegrow]

1 minute ago, TripleM999 said:

Now i see what they have done.

Heat Equalizing.sav 609.09 kB · 0 downloads

They completely removed the equalizing, not only the buggy temperature swap, but the heat neutral quick equalizing too. The fields, i painted in the save, would have the instant equalizing before the patch. Now they do something completely different, a complete tile swap, similar to the top-down-swap. This explains the more violent fluctuating gas behavior. With equalizing completely vanished, it exhibits the minuscule conductivity of gases more prominently.

Yeah, just like I said - "It's borked af"