Can we have a serious discussion about heat?

[mjl1966]

OK, so I've searched around for some info on this, and all I seem to find is "use moar radiators."  Well, that's nice, but is there a way to interpret the numbers so we can actually plan our loadouts?

My problem is not understanding the correlation between the power rating on the active devices and the radiators.  For example, the 1.25 ISRU converter requires 100 kW of cooling.  But if I strap on three 200 kW radiators, for a total of 600 kW of cooling, the converter still overheats in a matter of minutes if it has enough ore to run at full grunt.

Also curious is that I can keep one junior drill at around 650 degrees if I give it 600 kW of cooling even though it says it "requires" just 50.

Well, so, what's the deal here?  How can I calculate what to strap onto things to actually remove enough heat to keep them operational?  Obviously a kW of heat and a kW of cooling don't line up in the thermodynamics model.  (Which isn't saying it's wrong - I just don't understand it.)

Thanks.

 

[Jarin]

Agreed. I'd love to be able to have a direct correlation between listed stats in the VAB and functionality in space. Some idea of how to wring proper efficiency out of these systems without having to consult a wiki. 

[Reactordrone]

The radiator panels have to be very close to the heat producing part to have any effect. If they're not on a part directly attached to your ISRU or drill they might as well not be there at all.If you can't fit a radiator in an appropriate place you need to use the foldable thermal control panels. A single 200kW radiator will cool the large ISRU unit if it's in the right place.

Edited June 21, 2016 by Reactordrone

[Jarin]

As I understand it, extendible radiators can pull from up to... I think it's 3 parts away? Surface-attached passive cooling radiators only pull from the part they're attached to.

[fourfa]

Surface-attached radiators are 'supposed' to only cool the part they're attached to (the 'parent' part), and one part away ('sibling' part).  Before version 1.1, this mechanism did not really work.  From the 1.1 release changelog http://wiki.kerbalspaceprogram.com/wiki/1.1 -

• Fixed a regression where static radiators were not respecting their parent/sibling cooling limitations.

Extending radiators cool the entire vessel.  Their greater mass is explained by a system of coolant pipes running through the whole ship (saw that as an explanation by a dev on the pre-release subforum, which is no longer visible).  If you're having cooling trouble with ISRU, try a pair of medium folding radiators and try again.

[bewing]

An example of what fourfa said: I have a drill ship. At the bottom end it has a fuselage section, with 2 "large" surface mount radiators (200kW each), and 2 large drills directly attached to that fuselage section. This provides twice the cooling that the drills need, and they stay at 500 degrees the way they are supposed to.

I have an ISRU on a rover, with two deployable small radiators (50kW each). The rover attaches to the drill ship with a klaw and then starts the ISRU. The ISRU goes to 1000 degrees, and then slowly overheats, because that is not enough cooling, and the fixed panels on the drill ship only cool the nearby parts -- the ISRU is too far away.

So, beyond all that -- for the surface radiators, the "attached part plus every part directly attached to it" algorithm works correctly.

For the deployable radiators, there are some bugs in their heating model -- but I haven't been able to track them down completely. One point is that since they pull heat from the entire ship, if you want them to be pulling heat specifically from the ISRU, that's too darn bad. Every part in the ship seems to get equal weighting for how much heat gets pulled out. So the radiator might be pulling 25kW out of a drill that already has sufficient cooling, leaving 25kW of cooling power for the overheating ISRU that's right next to it.

[kiwi1960]

this topic is too hot for me, sorry.

[cantab]

There are three "sections" for heat and temperature on KSP parts. Skin represents the surface, internal represents the bulk of the part, and core represents the working parts of things like the ISRU converter. The "headline" rating on the radiators only applies to cooling skin and internal heat, and their ability to draw away core heat is significantly less.

[Jarin]

While the answers so far are generally good information, the fact remains that none of this is communicated in-game, and the wiki is several versions out of date.

[bewing]

I disagree. If you right click on the radiator part in the VAB, it says the kW rating and that surface mount panels must be close to the part that needs cooling. As far as other in-game info goes, that is what sandbox mode is for -- you have to test stuff to see how it works. As far as the wiki goes, it's the users' job to update it -- so if it's out of date it's our/your fault.

 

[Lunar Sea]

13 hours ago, fourfa said:

Surface-attached radiators are 'supposed' to only cool the part they're attached to (the 'parent' part), and one part away ('sibling' part).  Before version 1.1, this mechanism did not really work.  From the 1.1 release changelog http://wiki.kerbalspaceprogram.com/wiki/1.1 -

• Fixed a regression where static radiators were not respecting their parent/sibling cooling limitations.

Extending radiators cool the entire vessel.  Their greater mass is explained by a system of coolant pipes running through the whole ship (saw that as an explanation by a dev on the pre-release subforum, which is no longer visible).  If you're having cooling trouble with ISRU, try a pair of medium folding radiators and try again.

Ok, so I am running 1.0.5 and use 4 small radiator panels on a fuel tank adjacent to my big ISRU. Are you telling me they aren't doing squat? I haven't noticed any overheating issues. 

[fourfa]

No, I'm saying that in 1.0.5 they cool the whole ship. Very well in fact! A pair of medium surface rads always kept my ISRU ships very happy.

The same ship in 1.1 WILL overheat though, unless the drills and ISRU are sibling parts. 

[Jarin]

4 hours ago, bewing said:

I disagree. If you right click on the radiator part in the VAB, it says the kW rating and that surface mount panels must be close to the part that needs cooling. As far as other in-game info goes, that is what sandbox mode is for -- you have to test stuff to see how it works. As far as the wiki goes, it's the users' job to update it -- so if it's out of date it's our/your fault.

 

You're right, the real question is what sort of detail the KSPedia goes into. I had thought the wiki, being hosted by squad, would be the same info as KSPedia, but apparently I was mistaken. It looks like that's quite a bit more informative, so assuming there's no current bugs, I think I can withdraw my complaints.

[fourfa]

4 hours ago, bewing said:

I disagree. If you right click on the radiator part in the VAB, it says the kW rating and that surface mount panels must be close to the part that needs cooling. As far as other in-game info goes, that is what sandbox mode is for -- you have to test stuff to see how it works. As far as the wiki goes, it's the users' job to update it -- so if it's out of date it's our/your fault.

 

The new KSPedia entries on ISRU and cooling are more up to date, if I recall

[Geschosskopf]

On 6/20/2016 at 11:26 PM, mjl1966 said:

Well, so, what's the deal here?  How can I calculate what to strap onto things to actually remove enough heat to keep them operational?  Obviously a kW of heat and a kW of cooling don't line up in the thermodynamics model.  (Which isn't saying it's wrong - I just don't understand it.)

OK, to begin with, you don't need radiators at all except in 2 cases:

• Running something that produces lots of heat
  • ISRU and drills:  Work on the "Core" heat system
  • Nuclear engines (very unrealistically) work on the surface and internal system.
• Going well inside Moho's orbit, where the sunlight is strong enough to be a problem.  You don't need radiators just to orbit Moho.  So, there being very little reason to go inside of Moho, we'll ignore this one.

First off, the "core" heat of the ISRU and drills.  When you right-click on these parts, you'll see a line of text that says "cooling required" expressed in kW.  This tells you how much radiator capacity you need.  Then right-click on a radiator and look for the line that says "Core Heat Xfer", also expressed in kW.  This is the stat that matters when it comes to cooling ISRU and drills.  So, add up the "cooling required" values of all the ISRUs and drills you want to have running at once.  Then add up the "Core Heat Xfer" values of the radiators until that total equals or exceeds the "cooling required" total.  That's all there is to it.

With nuclear engines, there doesn't seem to be any easy math like with ISRU, so you just have to rely on rules of thumb.  A lot depends on what the engines are attached to and how long your burn them at what throttle setting.  An engine mounted radially on a skinny outrigger will get hot faster than if it's attached direction to a big fuel tank, for example.  But in general, 2 medium radiators will keep a single LV-N happy.  If you use the Atomic Age mod, that will also keep a LANTERN happy.  The small, RTG-based engines need 1 small radiator, and the big nuclear lightbulb needs 2 large radiators.

NOTE:  In all cases, I'm talking about retractable radiators.  Those are the only ones I use.  This is for 2 reasons:

• Fixed radiators only affect nearby parts.  This often causes problems mounting them where they'll be effective.  Also, if your heat-producing parts are spread out all over your ship, you end up needed more radiators in total than if you used retractable ones.
• Radiators don't work very well if the sun's shining on them.  t'he retractable ones pivot to stay edge-on to the sun but the fixed ones can only be reoriented by rotating the whole ship, which isn't always doable.  This means you need to use more radiators than really necessary so you'll always have enough radiators on the shady side of the ship to make up for those on the sunny side not working at full capacity.  And this complicates the issue of mounting the fixed radiators close enough to where you need them.

[Leftotian]

For example, the 1.25 ISRU converter requires 100 kW of cooling.  But if I strap on three 200 kW radiators, for a total of 600 kW of cooling, the converter still overheats in a matter of minutes if it has enough ore to run at full grunt.

 

Small ISRU needs 100kW/converter(up to 400), but "max cooling 50kW" so no matter how many radiators you use it will only get 50kW. That means that small ISRU can't not overheat. Just use one 10kg 50kW radiator directly attached to ISRU for max performance, you can't make it run better than that. 

Never use retractable radiators, they are bad and broken, which makes them twice as bad. For example, if you are trying to run full power large ISRU + 4 large drills you will need 2500kW = 2.5t of retractable radiators or just 900kW = 220 kg of surface attach radiators. Actually all radiators are broken(actually actually, drills are broken, but whatever), but with surface attach ones you can avoid the bug. You need to make sure that drill radiators don't interact with ISRU and vice versa, so make sure to put your drills/ISRU on different craft parts and keep your drills/ISRU radiators separated. 

[Geschosskopf]

6 hours ago, Leftotian said:

Never use retractable radiators, they are bad and broken,

Um, no, there is no bug.  Retractables are the only ones you can rely on.  They work no matter where you place them and you don't need to add extras due to the orientation of the ship to the sun.  With fixed radiators, you always need 25-50% more than the numbers alone say, due to shady and sunny sides of the ship, and you have to put them all on top of the heat-producing parts due to their limited reach.  Without inviting the Kraken from excessive clipping, fixed radiators become an exercise in futility fairly quickly.

[Leftotian]

24 minutes ago, Geschosskopf said:

Um, no, there is no bug.  Retractables are the only ones you can rely on.  They work no matter where you place them and you don't need to add extras due to the orientation of the ship to the sun.  With fixed radiators, you always need 25-50% more than the numbers alone say, due to shady and sunny sides of the ship, and you have to put them all on top of the heat-producing parts due to their limited reach.  Without inviting the Kraken from excessive clipping, fixed radiators become an exercise in futility fairly quickly.

Literally every single thing you said is wrong. 

There is a bug.

Retractible radiators are unreliable because of the bug.

The fact that they work no matter where you place them is what causing the bug and the very reason why you should not use them.

Fixed radiators do not rely on orientation of the sun, so you need to use the exact amount, while you need to use 200-300-400% more retractible radiators to compensate for the bug, which, combined with the fact, that they are 5 times heavier than fixed ones of same power, makes them weight 10 times more for same amount of cooling.

The fact that you need to put fixed radiators on same part as coolant consumer is an advantage, that makes them way better than retractible radiators.

Exercise in futility is using 4 tons of retractible radiators to coold down a craft, that will function just as well with 300 kg of fixed ones.

[Geschosskopf]

24 minutes ago, Leftotian said:

Literally every single thing you said is wrong. 

Hmm, so far you have not offered any evidence at all of a bug.  There is no bug that I know of associated with radiators.  Explain what the bug is and maybe I'll take you seriously.

Fixed radiators absolutely rely on orientation to the sun.  Radiators work by being hotter than their surroundings, so they can actually radiate heat into their surroundings.  If the sun is shining on them, they're hotter than if they're in the shade.  I mean, compare the temperatures experienced by stuff in low Earth orbit depending on if it's in daylight or Earth's shadow.  Thus, firxed radiators don't work well in sunshine, and the closer you are to the sun, the less well they work.  Of course, no radiator can help at all beyond the local ambient temperature, but the whole reason the retractable radiators rotate to stay edge-on to the sun is to eliminate the orientation problem.

Edited July 9, 2016 by Geschosskopf

[Leftotian]

What proof? Just go play the game and see for yourself.Test stuff right on the runway to save time. For example make large ISRU + 8 large drills. Run two channels, lf and ox(if you run more than 2 it will overheat no matter what, it can only take 500 and you need 200/channel).Test with retractible, then with fixed. Just don't forget to put ISRU/drills and their radiators on different parts. Tell me how much radiators you need to run it all without overheating. I can tell you right now how much, but try to figure out yourself. I can tell you that you will need 3x kW in retractibles, which means ~15x more weight.

[bewing]

@Leftotian: I did my version of your test. First, "broken" means "does not work, or works so badly that it is unusable". But all radiators provide cooling, and drills produce ore, so to claim that any of it is broken is BS. There are many gradiations of bugginess -- radiators are a little bit quirky, and drills are a little bit quirky ... but that's all I'm going to give you on that score.

The ship I used to test, and screenshots of the results are available on request -- but to summarize: there is indeed a bug, but it is in the Convert-o-tron 250, not in the radiators. The converter will overheat even when it has an overabundance of core cooling capacity from retractable radiators. It is true that it does not happen with fixed radiators. The way you can tell that this is a bug in the converter is that it does not affect drills. Drills reach their proper operating temp and hit a wall, the way they are supposed to.

I also did the test on Kerbin's runway at noon, and near sunset to test the solar aspect of @Geschosskopf's comments. I think I did notice an effect, but it is down in the 1% range for cooling efficiency. In space, the effect may be bigger, but I didn't test there. In any case, the solar aspect does not affect core cooling -- only internal and skin cooling.

Of course, there is always the deal about "which device consumes the electricity first" or "which jet engine consumes the air and fuel first" or, in this case, "which parts get cooled before all the other parts". That is an aspect of the game's engine and is not a bug in a particular part.

 

Edited July 9, 2016 by bewing

[Geschosskopf]

5 hours ago, bewing said:

I also did the test on Kerbin's runway at noon, and near sunset to test the solar aspect of @Geschosskopf's comments. I think I did notice an effect, but it is down in the 1% range for cooling efficiency. In space, the effect may be bigger, but I didn't test there. In any case, the solar aspect does not affect core cooling -- only internal and skin cooling.

The effect of the sun on radiators varies greatly depending on where you are.  It's more noticeable on Mun than Kerbin, for example, because Kerbin's atmosphere greatly assists the functioning of radiators.  The atmosphere allows heat to leave the radiators via conduction and convection as well as just radiation, which is the only mechanism of heat transfer in the vacuum.  The closer you are to the sun, bigger the difference you get, too.

And solar issues do indeed affect how radiators work with the core heat system.  This is because the radiators have to get rid of the heat they absorb from the other parts of the ship (whether those parts use the core or the skin/internal heat system).  Radiators get rid of the absorbed heat via the skin/internal mechanism, and the skin/internal mechanism, which is affected by sunlight.  See, heat only moves from hotter things to colder things, and the bigger the difference in temperature, the faster the heat flows.  Therefore, the hotter your radiators are, the less effectively they both absorb heat from other parts and dump it to the surroundings.  This can result in them providing insufficient cooling for the core heat-producing parts.

[bewing]

43 minutes ago, Geschosskopf said:

This is because the radiators have to get rid of the heat they absorb from the other parts of the ship (whether those parts use the core or the skin/internal heat system).

That is true, but a medium Thermal Control System can do almost 20MW of cooling, total -- with 99% of that being devoted specifically to skin and internal heat. Only a mere 250kW is devoted to core cooling. So for a radiator to not be able to suck out a full quota of core heat, it has to already be totally maxed out. In the testing I did, the radiators were handling a constant 16.5% load because of absorption of ambient heat -- but an absolute minimum number of radiators could still keep any drill or ISRU from overheating. (Provided you don't use both drills and an ISRU at the same time, because that starts giving slightly buggy behavior.)

Edited July 9, 2016 by bewing

[Geschosskopf]

22 minutes ago, bewing said:

but an absolute minimum number of radiators could still keep any drill or ISRU from overheating.

Hmm, good to know.  It used to not be that way.

[Leftotian]

9 hours ago, bewing said:

 

 there is indeed a bug, but it is in the Convert-o-tron 250, not in the radiators. The converter will overheat even when it has an overabundance of core cooling capacity from retractable radiators. It is true that it does not happen with fixed radiators. The way you can tell that this is a bug in the converter is that it does not affect drills. Drills reach their proper operating temp and hit a wall, the way they are supposed to.

And you are wrong. Not enough testing, no understanding of mechanics and a random guess on which part is bugged, based on converter overheating.

As i said already, drills are bugged. But that drill bug in practice makes radiators not work properly, which in turn overheats a converter.

Converter is not bugged, converter will not overheat if it has overabundance of cooling, converter will not overheat even when it has the exact amount of cooling. The problem is that it does not get this amount because of a drill bug, drills will consume up to 500%(actually up to infinity, if you provide that, but in practice it doesn't matter after 500%, at that point you will have enough coolant to feed every part in the game even with the bug) of their normal amount, taking away from a converter and overheating it. Dividing radiators in drill only and converter only by using surface radiators prevents drills from starving converter.

It's an old bug, it was here since at least 1.0.4. Actually in 1.0.4. and 1.0.5. converter was also buged in a same way, but it was fixed in 1.1.0. Drills were not.

Edited July 9, 2016 by Leftotian