how do radiator/active radiator modules work?

[Rushligh]

im trying to create a unique set of heat pumps and am having difficulty finding information on how the KSP radiator modules work.
 

It's easy enough to convert electricity into a resource to act as an abalator, (what one cooling systems pack does), but i'd like to know the radiators full inner workings etc.

[Streetwind]

A lot of that can be observed ingame if you hit Alt+F12 and go to physics -> thermal and check "show debug info in menus".

Very briefly, heat in KSP works like this:
 

Spoiler

 

- You have four different flux components, where "flux" simply means "something is flowing/being exchanged". There's conduction, which is heat transfer between adjacent parts across a temperature differential; there's convection, which is heat transfer between a part and an atmosphere or body of liquid; there's blackbody radiation, which is heat rejected into the great void of the vacuum; and there's internal flux, which describes parts generating or removing heat through their own internal functions, or those of nearby parts.

- Parts have a skin (which in turn can have two faces - I don't know how that works), some internals, and optionally a core. There is conduction flux between all three of those, the degree of which can be configured; also, conduction between parts happens on both the skin to skin level as well as the internals to internals level, but never between cores. A part's maxTemp field controls how hot both skin and internals can get without the part expoding; there's probably a way to set it individually too. The core always has its heat cap defined individually. If any of those maximum heat tolerances is exceeded, even for a split second, the part is instantly destroyed. Radiation and convection flux is exclusively happening from the skin.

- Temperature is given in Kelvin, heat energy is given in kilojoules, and heat flux is given in kilowatts.

- Skin, internals and core each have a "thermal mass". It governs how many Kelvin a section increases in temperature when a given amount of heat energy is added to it. Low thermal mass means a part increases in temperature very quickly. Skin has low thermal mass. This is why you need ablative heat shields.

- A passive radiator works by simply being a part with great conductive and radiative properties, which causes heat from adjacent parts to easily conduct into it and be radiated away. Note that all parts in the game passively radiate heat; specific passive radiators are just better at it than normal parts. However, Squad has already given all parts very generous self-radiating capabilities. Also, since conduction seeks to equalize temperatures between parts, a passive radiator can never cool a part below its own temperature. This makes them, by definition, the coldest part on the vessel - and they have to be, because otherwise, no heat would flow into them. Since blackbody radiation scales with the fourth power of the temperature, this makes passive radiators very slow, because they always remain at very low temperatures. Should environmental heat affect a passive radiator, they will actually ingest temperature more quickly than any other part, and conduct it into the spacecraft instead of cooling it. In stock KSP, the fixed radiator panels used to be passive radiators. However, due to the points described above, players found them extremely useless and they were made active in 1.0.5. Solar panels still function as passive radiators, however they do want to orient themselves towards the sun, thereby potentially ingesting heat instead of rejecting it. They also lose Ec production efficiency as their temperature increases.

- An active radiator works by checking other parts if they exceed a certain temperature threshold (for stock it's 400 K, I believe). If a part does exceed it, the radiator creates a negative internal flux in that part, and a positive internal flux of the same value in itself. Thus it steals the part's heat and transfers it to itself. Using this mechanic, an active radiator can become hotter than the part it wants to cool (in stock, up to four times as hot). This makes them by definition the hottest part on the vessel. Which is great, since blackbody radiation scales with the fourth power of the temperature. A very hot active radiator will reject a huge amount of heat, and will even continue to work (to some degree) in an environment that's nominally hotter than the vessel itself. All stock radiators currently are active radiators, and consume Ec/s to provide their function. Active radiators with a joint will attempt to orient themselves edge-on to the sun to minimize incoming heat from sunlight.

- Stock radiators have the ability to configure a field called "maxJumps" in their module, which defines their reach. The field defautls to 0, which makes a radiator operate vessel-wide, able to reach all parts. The extending stock radiators do this. The fixed panels, meanwhile, have a value of 2 max jumps set, meaning they can only reach the part they are attached to, and one jump out from that. This can have advantages or disadvantages, depending on what you want to do.

- Cycling back to "core" heat: As mentioned above, giving a part a core is optional. It is meant for parts which must generate heat during high timewarp, or passively in the background while the vessel is unfocused for extended periods of time. Due to the challenges involved in that, core heat "cheats" the system, and works separately from it. For starters, all active radiators have a maximum "core transfer" value, which is far below the radiator's theoretical capacity. For example, the largest stock radiator can radiate many dozens of megawatts effortlessly, but can only transfer one megawatt worth of core heat. Additionally, even that heat isn't transferred for real, but internally divided by 100 before transfer. So a large thermal control system fully tapping out its core transfer will show an internal flux of only 10 kilowatts. But because that isn't complicated enough, using the core transfer function of the radiator saturates some part of its normal (and usually much greater) heat transfer capability, and vice versa; a running rocket engine that produces enough heat to prompt the radiator into cooling it will mean that less than the full core transfer remains available to cool cores, even though the radiator is still cold and not anywhere near maxed out. Parts that use core heat are for example the drills and ISRU refineries; the system was introduced in 1.0.5 specifically for them, because they tended to get screwy under timewarp or straight-up blow up on load in previous versions.

 

Yes, that is the brief version I don't know all the details myself. Good luck in your endeavour...

Edited March 23, 2016 by Streetwind

[RoverDude]

Mostly correct - other than the purpose of core heat, which is to provide mechanics for a much wider range of internal temperatures that would blow up things if used as a normal part temp.  Consider a book sitting on top of your laptop.  The book gets warm because of the conductive heat from the laptop (this would be akin to the internal temp of the laptop bleeding into it's skin temp and transferring to the book).  Core temp would be like putting your finger directly on the CPU sans heat sink - i.e. a LOT hotter.  Another example (and closer to what we use for KSP) would be something like an arc furnace - which can cook at 2000-3000 degrees Kelvin (i.e. a temperature where a lot of parts get all glowy and start going poof!).  

Most of core heat is assumed to dissipate in exhaust and leavings, and is heavily insulated by the part itself.  Hence why it works on a completely different level than your basic conductivity/radiative/convective thermal stuff.  It affects it, and can be affected by it, but we have tools for managing it effectively, including overheating and thermal efficiency, without having to artificially lower it to avoid exploding your parts.

 

(source:  I wrote it ;))

Edited March 24, 2016 by RoverDude

[Rushligh]

hmm, that is all in theory correct. i do have 'deadly re-entry' installed as a requirement for KSPI: which changes the way heat is transferred and dissipated (in a very irritating way)

what I've been TRYING to do is make a powered inline cooling unit that consumes MW. using abalator no longer works on heat generated by parts (because of DR's heat mechanic)
aside of writing and compiling some c# scripts, is there a way to manage heat in this fashion? can a fuel have a temperature/conduct heat?

I've seen that exposing a part to an intake of cold air DOES make a difference. example:
Using Insanity's mk precooler part as an intake, i gave it the thermal transfer module for KSPI to act as a bridge for thermal engines. gave it an active radiator module, etc. i basically copied the inline radiator into that part and removed the intake module. it acted slightly better, however the body that the reactors were attached to over heated fast.

i then put the intake module back in, and not only sis the main body NOT overheat, the thermal engines took twice as long to over heat.

i tried this same thing with a very hot nuclear rocket in atmosphere and the results were the same.

[Booots]

Is there an easy way of tapping into the heat drawing mechanism of the active radiators? I'm thinking about making a geothermal cooling mod that would allow for much more powerful and efficient cooling while landed by using the surface as a heat sink. I'd rather not re-invent the wheel and code cooling logic from scratch if I could just make a child class of the active radiators and call some sort of draw heat method if that's possible instead.