Evaporation and condensation

[ThoroughbredFun]

its simple really, hot water would evaporate more  and cooled stuff would make condensation. 

 

I suggest adding a humidity factor to the game, making plants not be too humid.

also if there is anything but oxygen in the air the evaporated water would remove some of it and condense into contaminated water.

[Kasuha]

It would IMO allow implementing much smoother and accurate liquid/gas physics if the game supported atmosphere/liquids as mixes of components rather than per-tile bubbles. It would not prevent CO2 collecting at the bottom or hydrogen collecting at the top, gas/liquid filters and all machines could work almost the same how they work now and it would avoid all kinds of silly problems like a ton of water being blocked by a puddle with a few grams of contamination.

Sadly that's not how the game was implemented. I wish the devs reconsidered and went for reimplementation, I'm pretty sure it would be worth it.

Regarding evaporation, water could just evaporate at 100 C - any additional heat pumped to 100 C hot water just turns proportional amount of water to steam at 100 C, and any heat taken from steam at 100 C should turn proportional part of that steam to water instead of cooling it. I don't think we need humidity and condensation nuclei, just the basics would be fine.

[Moggles]

It's probably too intensive on CPU and memory to mix gasses and liquids.

Suddenly instead of having a tile object containing a single element and quantity you would have tile objects having to deal with an array of elements and quantities to resolve every time physics is updated. When you add more elements to the mix and think that there can be milligram quantities in a 2000g pressurised tile it would get out of hand fast.

Would be cool though. Are there any games around that attempted a simulation of that complexity?

[Kasuha]

2 hours ago, Moggles said:

It's probably too intensive on CPU and memory to mix gasses and liquids

Or it might be just too expensive (in terms of money) to implement it efficiently. I mean, processing arrays of cells (assuming we use fixed structures rather than dynamically allocated objects) is perfect task for vector instructions but few people bother with assembly language nowadays. I have no doubts it would fit in memory, restricting to bare data saves a lot of space otherwise used by pointers. CPU processing is a bit of question but I believe it would run at reasonable rate on modern CPUs. Computers are 1000 times faster now compared to times when I was doing comparable simulations in school (literally, I was using a 3 MHz machine and it was running real time, though on single "gas" and smaller grid).